Medical robot fault detection method and platform based on AI and storage medium
By constructing a multi-dimensional anomaly index and early warning level, the shortcomings of early fault detection of intravenous infusion pumps are addressed, enabling accurate early warning of infusion pumps and assessment of equipment health status, thereby improving infusion safety and management efficiency.
Patent Information
- Application Number
- CN202511768123.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-11-28
- Publication Date
- 2026-03-17
AI Technical Summary
Existing methods for detecting faults in intravenous infusion pumps rely on post-infection alarms or periodic manual inspections. These methods lack the ability to provide early warnings of performance degradation and latent faults, and cannot effectively distinguish between normal pressure changes caused by drug characteristics and mechanical failures. This poses clinical safety risks and management efficiency bottlenecks.
An AI-based fault detection method is adopted, which constructs pulse anomaly index, motor deviation index and comprehensive anomaly index, and combines drug characteristic compensation coefficient and thermal risk coefficient to generate early warning level, thereby realizing multi-dimensional perception and intelligent fusion analysis.
It enables precise early warning of early failures and accuracy degradation of intravenous infusion pumps, improves the specificity and accuracy of monitoring, transforms into pre-emptive prediction, ensures infusion safety, reduces unplanned downtime, and optimizes equipment management efficiency.
Smart Images
Figure CN121668445A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of artificial intelligence technology, and in particular to an AI-based method, platform, and storage medium for fault detection in medical robots. Background Technology
[0002] Currently, intravenous infusion pumps are widely used in clinical practice, but fault detection largely relies on post-infusion alarms or periodic manual inspections, lacking the ability to predict early performance degradation and latent faults. Traditional methods typically monitor only a single parameter, failing to effectively distinguish between normal pressure changes caused by drug characteristics and genuine mechanical failures, easily leading to false alarms or missed alarms.
[0003] Meanwhile, slow-onset precision degradation issues such as minor step loss in motors and slight wear on pump heads are difficult to detect before they lead to clinical accidents. This passive and isolated monitoring model cannot meet the needs of smart healthcare for predictive maintenance of equipment, and poses certain clinical safety risks and management efficiency bottlenecks. Therefore, there is an urgent need for a comprehensive fault detection solution that can perceive multiple dimensions, intelligently integrate and analyze data, and provide early warnings. Summary of the Invention
[0004] The purpose of this invention is to provide an AI-based method, platform, and storage medium for detecting faults in medical robots, in order to solve at least one of the problems existing in the prior art.
[0005] To achieve the above objectives, the present invention adopts the following technical solution: An AI-based fault detection method for medical robots includes: A pulse anomaly index is constructed based on flow pulse data and motor operation data, and a motor deviation index is constructed based on motor operation data. The comprehensive abnormality index is calculated by integrating the waveform abnormality index, pulse abnormality index and motor deviation index, and the drug characteristic compensation coefficient is constructed based on the viscosity and density of the infused drug to update the comprehensive abnormality index. A thermal risk coefficient is constructed based on the motor coil current and the motor casing temperature, and an early warning level is generated based on the comprehensive anomaly index and the thermal risk coefficient.
[0006] Furthermore, feature extraction is performed on the infusion tubing pressure within the analysis period to construct a waveform anomaly index: the infusion tubing pressure at the i-th sampling point of the j-th complete squeezing cycle within the analysis period is extracted as F(i,j), and the average infusion tubing pressure at the i-th sampling point of each complete squeezing cycle within the analysis period is calculated as Fpi, and the waveform anomaly index WCI is constructed.
[0007] Furthermore, a pulse anomaly index is constructed based on flow pulse data and motor operation data; The theoretical pulse count Pt is calculated based on the theoretical microstepping number Mt of the motor, the output volume Vp corresponding to each microstep of the pump tube, and the volume Vd of each drop of medicine. The pulse count Pa within the cycle is statistically analyzed. The pulse anomaly index PR is calculated based on the pulse count Pa and the theoretical pulse count Pt, where PR = |Pa - Pt| / Pt.
[0008] Furthermore, a motor deviation index is constructed based on motor operating data; Record the theoretical number of microsteps under each microstep command and the actual number of microsteps fed back by the driver chip within the analysis cycle. Take the absolute value of the difference between the theoretical number of microsteps under each microstep command and the actual number of microsteps fed back by the driver chip as the absolute deviation of each microstep. Calculate the average value AD of the absolute deviation of each microstep within the analysis cycle. Take the ratio of AD to the microstep deviation tolerance threshold as the motor deviation index PC.
[0009] Furthermore, based on the waveform anomaly index WCI, pulse anomaly index PR and motor deviation index PC within the analysis period, the comprehensive anomaly index CHS is calculated, CHS=w1×WCI+w2×PR+w3×PC, w1+w2+w3=1; Among them, w1 is the waveform anomaly weight, w2 is the pulse anomaly weight, and w3 is the motor deviation weight.
[0010] Furthermore, a viscosity factor is constructed based on the viscosity η of the infused drug and the viscosity threshold η0 during the analysis period. When η is less than or equal to η0, the viscosity factor is set to 0; otherwise, the viscosity factor is set to tanh[(η-η0) / η0]. A density factor is constructed based on the density ρ of the infused drug and the density threshold ρ0. When ρ is less than or equal to ρ0, the density factor is set to 0; otherwise, the density factor is set to {1-exp[-(ρ-ρ0) / ρ0]}. The drug property compensation coefficient DP is constructed based on viscosity factor and density factor, DP=x1×viscosity factor+x2×density factor, x1+x2=1, x1 is viscosity weight, x2 is density weight. The waveform abnormality index is updated based on the drug characteristic compensation coefficient DP, and the updated waveform abnormality index is set as WCI', WCI'=WCI×[1-min(DP,0.3)], to update the comprehensive abnormality index.
[0011] Furthermore, the average temperature of the motor housing within the analysis cycle is calculated as avg(T), and a temperature anomaly factor is constructed based on avg(T). When avg(T) is less than or equal to 0.9×Tm, the temperature anomaly factor is set to 0; when avg(T) is greater than 0.9×Tm but less than Tm, the temperature anomaly factor is set to 1-(Tm-avg(T)) / Tm; and when avg(T) is greater than or equal to Tm, the temperature anomaly factor is set to 1. The average value of the motor coil current during the analysis period is avg(I), and the standard deviation of the motor coil current during the analysis period is σ(I). A thermal risk coefficient is constructed based on avg(I), σ(I), and the temperature anomaly factor. The expression for the thermal risk coefficient is: TAI = α × [σ(I) / avg(I)] / B + β × temperature anomaly factor; Where B is the current fluctuation threshold, Tm is the maximum allowable operating temperature of the motor housing, α is the current fluctuation weight, β is the temperature anomaly weight, and α+β=1.
[0012] Furthermore, when the thermal risk coefficient TAI is less than or equal to the thermal risk threshold, if the comprehensive anomaly index is less than or equal to the first anomaly threshold, the warning level is determined to be normal; if the comprehensive anomaly index is greater than the first anomaly threshold and less than or equal to the second anomaly threshold, the warning level is determined to be a primary warning; if the comprehensive anomaly index is greater than the second anomaly threshold, the warning level is determined to be a high-level warning. When the thermal risk coefficient (TAI) is greater than the thermal risk threshold, the warning level is determined to be a high-level warning.
[0013] According to another aspect of this application, an AI-based medical robot fault detection platform is provided, comprising: The data acquisition unit is used to collect pressure, flow pulse data and motor operation data of the infusion pipeline; The waveform feature construction unit is used to extract features from the infusion tubing pressure during the analysis period in order to construct a waveform anomaly index. The feature extraction unit is used to construct a pulse anomaly index based on flow pulse data and motor operation data, and to construct a motor deviation index based on motor operation data. The fusion analysis unit is used to fuse waveform abnormality index, pulse abnormality index and motor deviation index to calculate comprehensive abnormality index, and to construct drug characteristic compensation coefficient based on the viscosity and density of the infused drug to update the comprehensive abnormality index. The early warning unit is used to construct a thermal risk coefficient based on the motor coil current and the motor casing temperature, and to generate an early warning level based on the comprehensive anomaly index and the thermal risk coefficient.
[0014] According to another aspect of this application, a computer-readable storage medium is provided, the computer-readable storage medium storing a computer program, wherein the computer program is used to control the electronic device in which the computer-readable storage medium is located to execute the AI-based medical robot fault detection method during runtime.
[0015] The beneficial effects of this invention are as follows: This invention enables accurate early warning of early failures and accuracy degradation of intravenous infusion pumps. The method comprehensively analyzes multi-dimensional characteristics such as pressure waveform consistency, flow pulse accuracy, motor execution fidelity, and thermal load risk. By introducing a drug characteristic compensation coefficient, it effectively eliminates the interference of different drug solution physical properties on fault judgment, significantly improving the specificity and accuracy of monitoring. Its ultimate advantage lies in transforming maintenance strategies from reactive to proactive, enabling timely detection of potential mechanical and electrical faults to ensure patient infusion safety. Furthermore, it allows for maintenance scheduling based on equipment health status, significantly reducing unplanned downtime and optimizing equipment management efficiency and maintenance costs in clinical departments. Attached Figure Description
[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0017] Figure 1 This is a flowchart illustrating the AI-based medical robot fault detection method in this embodiment.
[0018] Figure 2 This is a flowchart illustrating the feature construction method of this embodiment.
[0019] Figure 3 This is a flowchart illustrating the method for constructing the comprehensive anomaly index in this embodiment.
[0020] Figure 4 This is a schematic diagram of the AI-based medical robot fault detection platform in this embodiment. Detailed Implementation
[0021] To more clearly illustrate the present invention, the following description, in conjunction with preferred embodiments and accompanying drawings, further explains the invention. Similar components in the drawings are indicated by the same reference numerals. Those skilled in the art should understand that the specific description below is illustrative rather than restrictive and should not be construed as limiting the scope of protection of the present invention.
[0022] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate for the embodiments of this application described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.
[0023] Specifically, this embodiment is applied to early fault and accuracy degradation warning of intravenous infusion pumps in medical robots.
[0024] Please see Figure 1 As shown, it is a flowchart illustrating the AI-based medical robot fault detection method of this embodiment, including: Step S101: Collect pressure and flow pulse data of the infusion line and motor operation data. The flow pulse data is a pulse count. The motor operation data includes the theoretical microstep count, the actual microstep count, the motor coil current, and the motor housing temperature. The pulse count is the total number of drug drop pulses detected by the sensor within an analysis window, in drops. The theoretical microstep count is the total number of microsteps that the motor should complete within a certain time period, given by the control system command. The actual microstep count is the total number of microsteps actually completed within the same time period, fed back by the motor driver chip.
[0025] For example, in this embodiment, the pressure of the infusion line can be collected by a high-precision piezoelectric thin film sensor integrated in the infusion line, the flow pulse data can be collected by an infrared drip rate sensor, the theoretical microstep number, the actual microstep number, and the motor coil current can be read by the built-in register of the stepper motor driver chip of the infusion pump, and the temperature of the motor shell can be collected by a digital temperature sensor integrated in the pump body. This embodiment does not specifically limit the above settings, and those skilled in the art can freely set them according to their needs.
[0026] Specifically, a multi-source heterogeneous sensing system was used to simultaneously collect core raw data reflecting mechanical motion, fluid transport, and motor operation. This forms a data foundation for comprehensively understanding the working status of the infusion pump, ensuring that subsequent analysis can accurately characterize the equipment's operating status from multiple dimensions such as pressure, flow rate, control, and temperature, providing rich and reliable data for early fault detection.
[0027] Please continue reading. Figure 1 As shown, the AI-based medical robot fault detection method further includes: Step S102: Extract features from the infusion tubing pressure during the analysis period to construct a waveform anomaly index.
[0028] Specifically, the infusion tubing pressure at the i-th sampling point of the j-th complete squeezing cycle within the analysis period is extracted as F(i,j), and the average infusion tubing pressure at the i-th sampling point of each complete squeezing cycle within the analysis period is calculated as Fpi. A waveform anomaly index, WCI, is then constructed, and its expression is as follows: ; In the formula, I represents the number of sampling points within a complete extrusion cycle, and J represents the number of complete extrusion cycles within the analysis cycle.
[0029] Specifically, the complete squeezing cycle described in this embodiment is a complete mechanical action cycle that the pump head of the infusion pump goes through from the start of squeezing the tubing to its complete return to the starting position and preparation for the next squeezing. In the pressure sensor data, it is represented as a complete pressure waveform from the starting point of the waveform, through the peak value, and back to the starting point.
[0030] Specifically, the analysis period described in this embodiment is 10 minutes. This embodiment does not impose specific limitations on the setting of the analysis period, and those skilled in the art can set it freely according to their needs.
[0031] Specifically, by calculating the average deviation between the pressure waveform and the ideal waveform for each operating cycle, it is possible to sensitively detect minute anomalies in mechanical and fluid subsystems, such as pump head wear, pipeline aging, or blockage, thereby enabling early and quantitative assessment of the health status of mechanical transmission components.
[0032] Please continue reading. Figure 1 As shown, the AI-based medical robot fault detection method further includes: Step S103: Construct a pulse anomaly index based on flow pulse data and motor operation data, and construct a motor deviation index based on motor operation data.
[0033] Please see Figure 2 As shown, the feature construction method includes: Step S201: Construct a pulse anomaly index based on flow pulse data and motor operation data.
[0034] Specifically, the theoretical pulse count Pt is calculated based on the theoretical microstep number Mt of the motor, the output volume Vp corresponding to each microstep of the pump tube, and the volume Vd of each drop of medicine. Pt = (Mt × Vp) / Vd. The pulse count within the statistical analysis period is Pa, and the pulse anomaly index PR is calculated based on the pulse count Pa and the theoretical pulse count Pt, where PR = |Pa - Pt| / Pt.
[0035] Specifically, in this embodiment, the output volume Vp corresponding to each microstep of the pump tube and the volume Vd of each drop of medicine can be obtained through user interaction. The unit of Vp is (µL / microstep) and the unit of Vd is (µL / drop).
[0036] Specifically, by comparing the theoretically calculated number of droplets with the actual number of droplets monitored, the deviation in delivery accuracy is directly quantified. This index can effectively reveal inaccurate flow rates caused by pump head slippage, pipeline leakage, sensor contamination, or air in the infusion tubing, directly ensuring the safety and accuracy of clinical drug administration.
[0037] Please continue reading. Figure 2 As shown, the feature construction method further includes: Step S202: Construct a motor deviation index based on motor operating data.
[0038] Specifically, the theoretical number of microsteps under each microstep command and the actual number of microsteps fed back by the driver chip are recorded within the analysis cycle. The absolute value of the difference between the theoretical number of microsteps under each microstep command and the actual number of microsteps fed back by the driver chip is taken as the absolute deviation of each microstep. The average value AD of the absolute deviation of each microstep within the analysis cycle is calculated, and the ratio of AD to the microstep deviation tolerance threshold is taken as the motor deviation index PC.
[0039] Specifically, in this embodiment, the microstep deviation tolerance threshold is 10 microsteps.
[0040] Specifically, by monitoring the faithfulness of the motor in executing control commands, this index reflects whether the motor has potential problems such as missing steps, overshoot, or jamming. It can detect electrical and mechanical faults hidden behind the control signals, such as motor aging, drive circuit failure, or excessive mechanical resistance, providing a deep insight into the health status of the actuator.
[0041] Please continue reading. Figure 1 As shown, the AI-based medical robot fault detection method further includes: Step S104: The waveform abnormality index, pulse abnormality index and motor deviation index are combined to calculate the comprehensive abnormality index, and a drug characteristic compensation coefficient is constructed based on the viscosity and density of the infused drug to update the comprehensive abnormality index.
[0042] Please see Figure 3 As shown, the method for constructing the comprehensive anomaly index includes: Step S301: Calculate the comprehensive abnormal index by integrating the waveform abnormality index, pulse abnormality index, and motor deviation index.
[0043] Specifically, the comprehensive anomaly index CHS is calculated based on the waveform anomaly index WCI, pulse anomaly index PR and motor deviation index PC within the analysis period, CHS=w1×WCI+w2×PR+w3×PC, w1+w2+w3=1; Among them, w1 is the waveform anomaly weight, w2 is the pulse anomaly weight, and w3 is the motor deviation weight.
[0044] Specifically, in this embodiment, the waveform anomaly weight is 0.4, the pulse anomaly weight is 0.3, and the motor deviation weight is 0.3.
[0045] Specifically, this involves integrating three indices reflecting mechanical, flow, and electrical anomalies into a unified comprehensive health score. This integration provides a holistic perspective, avoiding the interference of false alarms from a single indicator, and making the assessment of the overall equipment status more comprehensive, stable, and reliable.
[0046] Please continue reading. Figure 3 As shown, the method for constructing the comprehensive anomaly index includes: Step S302: Construct a drug characteristic compensation coefficient based on the viscosity and density of the infused drug to update the comprehensive abnormality index.
[0047] Specifically, a viscosity factor is constructed based on the viscosity η of the infused drug and the viscosity threshold η0 during the analysis period. When η is less than or equal to η0, the viscosity factor is set to 0; otherwise, the viscosity factor is set to tanh[(η-η0) / η0]. A density factor is constructed based on the density ρ of the infused drug and the density threshold ρ0. When ρ is less than or equal to ρ0, the density factor is set to 0; otherwise, the density factor is set to {1-exp[-(ρ-ρ0) / ρ0]}. The drug property compensation coefficient DP is constructed based on viscosity factor and density factor, DP=x1×viscosity factor+x2×density factor, x1+x2=1, x1 is viscosity weight, x2 is density weight. The waveform abnormality index is updated based on the drug characteristic compensation coefficient DP, and the updated waveform abnormality index is set as WCI', WCI'=WCI×[1-min(DP,0.3)], to update the comprehensive abnormality index.
[0048] Specifically, in this embodiment, the viscosity threshold and density threshold are the viscosity (1.0 mPa·s) and density (1000 kg / m³) of 0.9% sodium chloride injection. The viscosity and density of the infused drug are obtained through an internally pre-set drug physical property database, which stores the drug's viscosity (in millipascals·seconds) and drug density (in kilograms per cubic meter).
[0049] Specifically, in this embodiment, the viscosity weight is 0.7 and the density weight is 0.3.
[0050] Specifically, if the infused drugs are not unique within the current analysis period, the comprehensive abnormality index will not be updated.
[0051] Specifically, by identifying the characteristic that high-viscosity, high-density drugs naturally lead to increased pipeline pressure and dynamically compensating for related indices, false alarms caused by different drugs are greatly reduced, and the accuracy and practicality of the algorithm in different medication scenarios are significantly improved.
[0052] Please continue reading. Figure 1 As shown, the AI-based medical robot fault detection method further includes: Step S105: Construct a thermal risk coefficient based on the motor coil current and the motor casing temperature, and generate an early warning level based on the comprehensive anomaly index and the thermal risk coefficient.
[0053] Specifically, the average temperature of the motor housing within the analysis period is calculated as avg(T), and a temperature anomaly factor is constructed based on avg(T). When avg(T) is less than or equal to 0.9×Tm, the temperature anomaly factor is set to 0; when avg(T) is greater than 0.9×Tm but less than Tm, the temperature anomaly factor is set to 1-(Tm-avg(T)) / Tm; and when avg(T) is greater than or equal to Tm, the temperature anomaly factor is set to 1. The average value of the motor coil current during the analysis period is avg(I), and the standard deviation of the motor coil current during the analysis period is σ(I). A thermal risk coefficient is constructed based on avg(I), σ(I), and the temperature anomaly factor. The expression for the thermal risk coefficient is: TAI = α × [σ(I) / avg(I)] / B + β × temperature anomaly factor; Where B is the current fluctuation threshold, Tm is the maximum allowable operating temperature of the motor housing, α is the current fluctuation weight, β is the temperature anomaly weight, and α+β=1. When the thermal risk coefficient (TAI) is less than or equal to the thermal risk threshold, if the comprehensive anomaly index is less than or equal to the first anomaly threshold, the warning level is determined to be normal; if the comprehensive anomaly index is greater than the first anomaly threshold and less than or equal to the second anomaly threshold, the warning level is determined to be a primary warning; if the comprehensive anomaly index is greater than the second anomaly threshold, the warning level is determined to be a high-level warning. When the thermal risk coefficient (TAI) is greater than the thermal risk threshold, the warning level is determined to be a high-level warning.
[0054] Specifically, the current fluctuation threshold in this embodiment ranges from [0.2, 0.3], and in this embodiment, the current fluctuation threshold is 0.25, the current fluctuation weight is 0.4, and the temperature anomaly weight is 0.6.
[0055] Specifically, in this embodiment, the thermal risk threshold is 0.6, the first abnormal threshold is 0.3, and the second abnormal threshold is 0.5.
[0056] Specifically, it transforms complex data indicators into clear and easy-to-understand warning levels, and innovatively integrates real-time abnormal states with long-term thermal risks. It not only focuses on current abnormal scores but also assesses the cumulative thermal risk of continuous motor overload, thus achieving dual protection that addresses both immediate faults and chronic degradation risks. The resulting risk levels provide healthcare professionals with clear, tiered guidance for handling such situations, optimizing maintenance processes.
[0057] Please see Figure 4 As shown, the AI-based medical robot fault detection platform includes: The data acquisition unit is used to collect pressure, flow pulse data and motor operation data of the infusion pipeline; The waveform feature construction unit is used to extract features from the infusion tubing pressure during the analysis period in order to construct a waveform anomaly index. The feature extraction unit is used to construct a pulse anomaly index based on flow pulse data and motor operation data, and to construct a motor deviation index based on motor operation data. The fusion analysis unit is used to fuse waveform abnormality index, pulse abnormality index and motor deviation index to calculate comprehensive abnormality index, and to construct drug characteristic compensation coefficient based on the viscosity and density of the infused drug to update the comprehensive abnormality index. The early warning unit is used to construct a thermal risk coefficient based on the motor coil current and the motor casing temperature, and to generate an early warning level based on the comprehensive anomaly index and the thermal risk coefficient.
[0058] The AI-based medical robot fault detection platform provided in this application can execute the AI-based medical robot fault detection method provided in any embodiment of this application, and has the corresponding functional modules and beneficial effects of the execution method.
[0059] This application also provides a computer-readable storage medium, which is a tangible physical storage medium that can store the aforementioned computer program and various types of data used in the program; the physical storage medium includes, but is not limited to, existing physical storage media or combinations thereof, such as random access memory, read-only memory, optical disk, and hard disk.
[0060] Those skilled in the art will understand that all or some of the steps and systems in the methods disclosed above can be implemented as software, firmware, hardware, and suitable combinations thereof. Some or all of the physical components can be implemented as software executed by a processor, such as a central processing unit, digital signal processor, or microprocessor, or as hardware, or as an integrated circuit, such as an application-specific integrated circuit. Such software can be distributed on a computer-readable medium, which can include computer storage media (or non-transitory media) and communication media (or transient media). As is known to those skilled in the art, the term computer storage media includes volatile and non-volatile, removable and non-removable media implemented in any method or technique for storing information (such as computer-readable programs, data structures, program modules, or other data). Furthermore, it is known to those skilled in the art that communication media typically contain computer-readable programs, data structures, program modules, or other data in modulated data signals such as carrier waves or other transmission mechanisms, and can include any information delivery medium.
[0061] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not intended to limit the implementation of the present invention. For those skilled in the art, other variations or modifications can be made based on the above description. It is impossible to exhaustively list all the implementation methods here. All obvious variations or modifications derived from the technical solutions of the present invention are still within the protection scope of the present invention.
Claims
1. An AI-based fault detection method for medical robots, characterized in that, The method comprises the following steps: Based on the flow pulse data and the motor operation data, a pulse anomaly index is constructed, and a motor deviation index is constructed based on the motor operation data; Fusion waveform anomaly index, pulse anomaly index and motor deviation index to calculate the comprehensive anomaly index, and according to the viscosity and density of the infusion drug to construct the drug characteristic compensation coefficient to update the comprehensive anomaly index; Based on the motor coil current and the motor shell temperature, a thermal risk coefficient is constructed, and a warning level is generated according to the comprehensive anomaly index and the thermal risk coefficient. 2.The AI-based medical robot failure detection method of claim 1, wherein The pressure of the infusion pipeline in the analysis period is extracted to construct the waveform anomaly index: the pressure of the infusion pipeline at the i-th sampling point of the j-th complete compression cycle in the analysis period is extracted as F(i,j), and the average value of the infusion pipeline pressure at the i-th sampling point of each complete compression cycle in the analysis period is calculated as Fpi, and the waveform anomaly index WCI is constructed. 3.The AI-based medical robot failure detection method of claim 2, wherein, Based on the flow pulse data and the motor operation data, a pulse anomaly index is constructed; According to the theoretical micro-step number Mt of the motor, the output volume Vp corresponding to each micro-step of the pump pipe, and the volume Vd of each drop of liquid, the theoretical pulse count Pt is calculated, and the pulse count in the analysis period is analyzed as Pa, and the pulse anomaly index PR is calculated based on the pulse count Pa and the theoretical pulse count Pt, PR=|Pa-Pt| / Pt. 4.The AI-based medical robot failure detection method of claim 3, wherein, Based on the motor operation data, a motor deviation index is constructed; The theoretical micro-step number and the actual micro-step number fed back by the driving chip under each micro-step instruction in the analysis period are recorded, and the absolute value of the difference between the theoretical micro-step number and the actual micro-step number under each micro-step instruction is taken as the absolute deviation of each micro-step, and the average value AD of the absolute deviation of each micro-step in the analysis period is calculated, and the ratio of AD to the micro-step deviation tolerance threshold is taken as the motor deviation index PC. 5.The AI-based medical robot failure detection method of claim 4, wherein, Based on the waveform anomaly index WCI, the pulse anomaly index PR and the motor deviation index PC in the analysis period, the comprehensive anomaly index CHS is calculated, CHS=w1×WCI+w2×PR+w3×PC, w1+w2+w3=1; Wherein, w1 is the waveform anomaly weight, w2 is the pulse anomaly weight, and w3 is the motor deviation weight. 6.The AI-based medical robot failure detection method of claim 1, wherein According to the viscosity η of the infusion drug in the analysis period and the viscosity threshold η0, a viscosity factor is constructed, when η is less than or equal to η0, the viscosity factor is set to 0, otherwise, the viscosity factor is set to tanh[(η-η0) / η0]; According to the density ρ of the infusion drug and the density threshold ρ0, a density factor is constructed, when ρ is less than or equal to ρ0, the density factor is set to 0, otherwise, the density factor is set to {1-exp[-(ρ-ρ0) / ρ0]}; Based on the viscosity factor and the density factor, a drug characteristic compensation coefficient DP is constructed, DP=x1×viscosity factor+x2×density factor, x1+x2=1, x1 is the viscosity weight, and x2 is the density weight; The waveform anomaly index is updated based on the drug characteristic compensation coefficient DP, and the updated waveform anomaly index is set as WCI', WCI'=WCI×[1-min(DP,0.3)], to update the comprehensive anomaly index. 7.The AI-based medical robot failure detection method of claim 6, wherein, An average value of the motor housing temperature in the analysis period is calculated as avg(T), and a temperature anomaly factor is constructed based on avg(T), the temperature anomaly factor is set to 0 when avg(T) is less than or equal to 0.9*Tm, the temperature anomaly factor is set to 1-(Tm-avg(T)) / Tm when avg(T) is greater than 0.9*Tm and less than Tm, and the temperature anomaly factor is set to 1 when avg(T) is greater than or equal to Tm; An average value of the motor coil current in the analysis period is calculated as avg(I), a standard deviation of the motor coil current in the analysis period is calculated as σ(I), and a thermal risk coefficient is constructed based on avg(I), σ(I), and the temperature anomaly factor, the expression of the thermal risk coefficient is: TAI=α×[σ(I) / avg(I)] / B+β×temperature anomaly factor; wherein B is a current fluctuation threshold, Tm is the maximum allowable operating temperature of the motor housing, α is a current fluctuation weight, β is a temperature anomaly weight, and α+β=1. 8.The AI-based medical robot failure detection method of claim 7, wherein, When the thermal risk coefficient TAI is less than or equal to a thermal risk threshold, if the comprehensive anomaly index is less than or equal to a first anomaly threshold, it is determined that the warning level is normal, if the comprehensive anomaly index is greater than the first anomaly threshold and less than or equal to a second anomaly threshold, it is determined that the warning level is primary warning, and if the comprehensive anomaly index is greater than the second anomaly threshold, it is determined that the warning level is advanced warning. When the thermal risk coefficient TAI is greater than the thermal risk threshold, it is determined that the warning level is advanced warning.
9. An AI-based medical robot failure detection platform applied to the AI-based medical robot failure detection method according to any one of claims 1-8, characterized in that, The method comprises: a data acquisition unit configured to acquire infusion pipeline pressure, flow pulse data, and motor operation data; a waveform feature construction unit configured to extract features from the infusion pipeline pressure in the analysis period to construct a waveform anomaly index; a feature extraction unit configured to construct a pulse anomaly index based on the flow pulse data and the motor operation data, and construct a motor deviation index based on the motor operation data; a fusion analysis unit configured to fuse the waveform anomaly index, the pulse anomaly index, and the motor deviation index to calculate a comprehensive anomaly index, and construct a drug property compensation coefficient based on the viscosity and density of the infusion drug to update the comprehensive anomaly index; a warning unit configured to construct a thermal risk coefficient based on the motor coil current and the motor housing temperature, and generate a warning level based on the comprehensive anomaly index and the thermal risk coefficient.
10. A computer-readable storage medium, characterized in that, The computer readable storage medium stores a computer program, wherein the computer program is used to control the electronic device where the computer readable storage medium is located to execute the AI-based medical robot fault detection method of any one of claims 1-8 when running. The computer readable storage medium stores a computer program, wherein the computer program is used to control the electronic device where the computer readable storage medium is located to execute the AI-based medical robot fault detection method of any one of claims 1-8 when running.