A graded fault-tolerant monitoring method and related equipment for motor temperature sensors

By using a data-driven model to perform graded fault-tolerant monitoring of motor temperature sensors, the problem of insufficient sensor data reliability assessment in existing technologies is solved. This enables refined judgment and graded response of motor operating status, improving the reliability and safety of the system.

CN122863241BActive Publication Date: 2026-11-17JIHUA LAB
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202611374965.6
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2026-09-07
Publication Date
2026-11-17
Estimated Expiration
2046-09-07

AI Technical Summary

Technical Problem

Existing technologies lack a mechanism for assessing the reliability of temperature sensor data in motor drive systems, making it impossible to finely process the gradual process of a sensor going from suspicious to confirmed failure. Furthermore, the lack of dynamic tracking and graded response leads to misjudgments and unplanned downtime.

Method used

By periodically acquiring the motor's temperature and operating data, a pre-trained data-driven model is used to predict the temperature and quantify the prediction error. By combining confidence intervals and joint fault probabilities, graded fault-tolerant monitoring is performed to achieve reliability assessment and graded response of the main temperature sensor.

Benefits of technology

It improves the reliability and safety of temperature monitoring, reduces false alarms and unplanned downtime, and enables redundant temperature prediction and hierarchical fault-tolerant control for sensor anomalies.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122863241B_ABST
    Figure CN122863241B_ABST
Patent Text Reader

Abstract

This invention discloses a graded fault-tolerant monitoring method and related equipment for motor temperature sensors. In the field of data processing technology, the method involves inputting auxiliary monitoring temperature and operating parameters into a pre-trained data-driven model to obtain a predicted temperature, which is then compared with the main sensor temperature to obtain the prediction error. The reliability of the main temperature sensor is determined by comparing the predicted error with a confidence interval obtained from historical prediction error statistics, thereby achieving a reliability assessment of the main sensor temperature and avoiding misjudgments caused by sensor distortion. In a reliable state, the main sensor temperature is compared with a preset high-temperature safety threshold to determine whether it is normal or faulty. In an unreliable state, the combined fault probability is quantified based on the predicted temperature, the preset high-temperature safety threshold, the operating error, and error statistics parameters, and compared with a preset probability judgment threshold to determine a warning or fault, thus enabling graded responses. This invention can reduce false alarms and unplanned downtime, improving the reliability and safety of temperature monitoring.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a graded fault-tolerant monitoring method and related equipment for motor temperature sensors. Background Technology

[0002] In highly integrated applications such as motor drive systems, especially robotic integrated joints, temperature monitoring is a crucial element in ensuring the safe operation of equipment. Currently, the following technical solutions are mainly used in this field: Physical contact temperature sensor solution: Negative temperature coefficient thermistors or platinum resistance temperature sensors are installed at critical locations such as the motor stator windings or power module substrates, reflecting temperature through changes in resistance. This solution can only detect hardware faults in the sensor circuit itself and lacks a mechanism for evaluating the reliability of the sensor data. Even if the sensor does not have a hard fault, its readings may be distorted due to aging, loosening, or local hot spot deviation.

[0003] The multi-sensor redundancy voting scheme uses multiple temperature sensors connected in parallel to sort the collected temperatures, calculate the differences, and then output either the average or an alarm based on these differences. However, its processing logic is binary: either it operates normally or it switches models. It lacks fine-grained processing capabilities in the ambiguous state of sensor readings being "suspicious but not completely ineffective," and after switching to pure model estimation, it cannot use historical sensor data to calibrate the model.

[0004] Model-referenced temperature estimation schemes indirectly estimate temperature by identifying the stator resistance of the motor online, serving as a backup in case of sensor failure. However, the accuracy of model estimation is greatly affected by changes in operating conditions, and existing schemes mostly adopt a passive switching mode that only activates the model after sensor failure, lacking dynamic tracking and graded response to the gradual process of sensor failure from "suspect" to "confirmed failure".

[0005] Data-driven fault diagnosis solutions: focus on fault detection itself. Summary of the Invention

[0006] The main objective of this invention is to provide a graded fault-tolerant monitoring method, device, electronic equipment, storage medium, and program product for motor temperature sensors, aiming to solve at least one problem in the prior art.

[0007] To achieve the above objectives, one aspect of the present invention proposes a graded fault-tolerant monitoring method for motor temperature sensors, the method comprising: The motor's temperature and operating data are acquired periodically. The temperature data includes the main sensing temperature collected by the main temperature sensor and the auxiliary monitoring temperature collected by the auxiliary temperature monitoring point. The operating data includes operating parameters and operating errors. The auxiliary monitoring temperature and operating parameters are input into a pre-trained data-driven model to predict the temperature. The prediction error is obtained by quantifying the main sensor temperature and the predicted temperature. The prediction error is compared with a preset confidence interval. If the prediction error falls within the confidence interval, the main temperature sensor is determined to be in a reliable state; otherwise, the main temperature sensor is determined to be in an unreliable state. The confidence interval is determined based on the mean and variance of the prediction error obtained from the statistical analysis of the historical prediction errors corresponding to several historical cycle nodes when the motor is in a normal state. When the main temperature sensor is in a reliable state, if the main sensor temperature does not exceed the preset high temperature safety threshold, the motor's operating state is determined to be normal; otherwise, the motor's operating state is determined to be faulty. When the main temperature sensor is in an unreliable state, the joint fault probability is quantified based on the predicted temperature, the preset high temperature safety threshold, the operating error, and the preset error statistics parameters. If the joint fault probability is less than the preset probability judgment threshold, the motor's operating state is determined to be a warning state; otherwise, the motor's operating state is determined to be a fault state. Based on the operating status, hierarchical responses are performed using preset mapping relationships.

[0008] In some optional methods, obtaining the motor's temperature data and operating parameters includes the following steps: The main sensing temperature is acquired by a main temperature sensor installed at a preset position on the winding of the motor; wherein, the main temperature sensor includes a negative temperature coefficient thermistor or a platinum resistance temperature sensor. At least one of the following temperatures—the power module board temperature, the motor housing temperature, and the driver ambient temperature—is collected by temperature sensors pre-installed on the power module board, housing, and driver of the motor as auxiliary monitoring temperatures. The motor's speed, power, and phase current are collected as operating parameters; Collect the command operation data and actual operation data of the joint module corresponding to the motor, and determine the operation error based on the difference between the command operation data and the actual operation data; The instruction execution data includes the instruction values ​​for the position, speed, and current of the joint module, while the actual execution data includes the actual values ​​for the position, speed, and current of the joint module.

[0009] In some alternative approaches, the method may also include the following steps: Acquire training data; where each training sample of the training data includes the historical main sensor temperature, historical auxiliary monitoring temperature and historical operating parameters of the same historical cycle node when the motor is in normal state, and different training samples of the training data correspond to different historical cycle nodes. Historical auxiliary monitoring temperatures and historical operating parameters are used as input data, and historical main sensor temperatures are used as label data to train the preset model for temperature prediction. In response to the results of temperature prediction training, the preset model is optimized and its parameters are tuned to obtain the trained model. The training model is validated using preset validation data to obtain validation results. If the accuracy of the validation results meets the preset accuracy conditions, the training model is used as the data-driven model. Otherwise, the training model is used as the preset model, and the process returns to the step of using historical auxiliary monitoring temperature and historical operating parameters as input data until the accuracy of the validation results meets the preset accuracy conditions. The preset models include neural network models, support vector regression models, or ensemble learning models.

[0010] In some alternative approaches, the method may also include the following steps: Multiply the prediction error variance by a preset confidence coefficient to obtain the boundary redundancy. The upper boundary is obtained by adding the mean of the prediction error to the boundary redundancy. Subtract the boundary redundancy from the mean prediction error to obtain the lower boundary; Determine the confidence interval based on the upper and lower boundaries.

[0011] In some alternative approaches, the error statistics parameters include the mean and variance of the operating error. When the main temperature sensor is in a reliable state, the method further includes the following steps: If the current cycle node is the first cycle node, the running error is used as the mean running error, and the running error variance is configured based on the preset initial value; If the current cycle node is not the first cycle node, update the error statistics parameters using the running error, including the following steps: The mean of the operating error of the error statistics parameter is used as the old mean of the operating error, and the variance of the operating error of the error statistics parameter is used as the old variance of the operating error. The first coefficient is obtained by using the preset number of times as the numerator and the sum of the preset number of times and 1 as the denominator; where the preset number of times represents the number of times the error statistics parameter is updated. Using 1 as the numerator and the sum of the preset number and 1 as the denominator, the second coefficient is obtained by calculation; The new mean operating error is obtained by adding the product of the first coefficient and the old mean operating error to the product of the second coefficient and the operating error. Multiply the square of the old operating error variance by the first coefficient to obtain the first value; The second value is obtained by multiplying the square of the difference between the operating error and the mean of the new operating error by the second coefficient. The square root of the sum of the first and second values ​​is taken to obtain the new operating error variance; Update the error statistics parameters using the new mean and variance of the operating error.

[0012] In some optional approaches, the error statistics parameters include the mean and variance of the operating error. The joint failure probability is quantified based on the predicted temperature, a preset high-temperature safety threshold, the operating error, and the preset error statistics parameters, including the following steps: Subtract the preset high-temperature safety threshold from the predicted temperature, and then add the prediction error variance to obtain the third value; Multiply the third value by the preset confidence coefficient to obtain the fourth value; The fifth value is obtained by taking the negative of the fourth value as the exponent of the natural constant. The sixth preset value is used as the numerator, and the sum of the fifth value and 1 is used as the denominator to calculate the temperature failure probability. The difference between 1 and the temperature failure probability is used as the normal temperature probability. The seventh value is obtained by taking the square of the difference between the running error and the mean running error as the numerator and twice the square of the running error variance as the denominator. The eighth value is obtained by using the natural exponential function based on the negative value of the seventh value. Multiply the eighth value by the preset ninth value to obtain the probability of normal operation, and take the difference between 1 and the probability of normal operation as the probability of operation failure. The product of the operational failure probability and the temperature failure probability is used as the numerator, and the product of the normal operation probability and the temperature normal operation probability plus the product of the operational failure probability and the temperature failure probability is used as the denominator to calculate the joint failure probability.

[0013] In some alternative approaches, a tiered response is performed based on the runtime status using a preset mapping relationship, including the following steps: When the operating status is normal, maintain the normal operation of the motor; When the operating status is in the warning state, maintain the normal operation of the motor and issue an abnormality reminder for the main temperature sensor; If the number of consecutive times the operating status is in the warning state reaches a preset threshold, the motor will be slowed down and stopped. When the operating status is faulty, the motor is slowed down and stopped.

[0014] To achieve the above objectives, another aspect of the present invention provides a graded fault-tolerant monitoring device for a motor temperature sensor, the device comprising: The data acquisition module is used to periodically acquire the motor's temperature data and operating data. The temperature data includes the main sensing temperature collected by the main temperature sensor and the auxiliary monitoring temperature collected by the auxiliary temperature monitoring point. The operating data includes operating parameters and operating errors. The temperature prediction module is used to input the auxiliary monitoring temperature and operating parameters into a pre-trained data-driven model to predict the temperature, and to obtain the prediction error based on the main sensor temperature and the predicted temperature. The sensing determination module is used to compare the prediction error with the preset confidence interval. When the prediction error falls within the confidence interval, the main temperature sensor is determined to be in a reliable state; otherwise, the main temperature sensor is determined to be in an unreliable state. The confidence interval is determined based on the mean and variance of the prediction error obtained from the statistical analysis of the historical prediction errors corresponding to several historical cycle nodes when the motor is in a normal state. The temperature determination module is used to determine the motor's operating status as normal if the main temperature sensor is in a reliable state and the main sensor temperature does not exceed the preset high temperature safety threshold; otherwise, it determines the motor's operating status as faulty. The probability determination module is used to quantify the joint fault probability based on the predicted temperature, the preset high temperature safety threshold, the operating error, and the preset error statistics parameters when the main temperature sensor is in an unreliable state. If the joint fault probability is less than the preset probability determination threshold, the motor's operating state is determined to be a warning state; otherwise, the motor's operating state is determined to be a fault state. The hierarchical response module is used to perform hierarchical responses based on the running status and using preset mapping relationships.

[0015] In some alternative embodiments, the apparatus also includes a model training module for performing the following operations: Acquire training data; where each training sample of the training data includes the historical main sensor temperature, historical auxiliary monitoring temperature and historical operating parameters of the same historical cycle node when the motor is in normal state, and different training samples of the training data correspond to different historical cycle nodes. Historical auxiliary monitoring temperatures and historical operating parameters are used as input data, and historical main sensor temperatures are used as label data to train the preset model for temperature prediction. In response to the results of temperature prediction training, the preset model is optimized and its parameters are tuned to obtain the trained model. The training model is validated using preset validation data to obtain validation results. If the accuracy of the validation results meets the preset accuracy conditions, the training model is used as the data-driven model. Otherwise, the training model is used as the preset model, and the process returns to the step of using historical auxiliary monitoring temperature and historical operating parameters as input data until the accuracy of the validation results meets the preset accuracy conditions. The preset models include neural network models, support vector regression models, or ensemble learning models.

[0016] In some alternative embodiments, the device also includes an interval statistics module for performing the following operations: Multiply the prediction error variance by a preset confidence coefficient to obtain the boundary redundancy. The upper boundary is obtained by adding the mean of the prediction error to the boundary redundancy. Subtract the boundary redundancy from the mean prediction error to obtain the lower boundary; Determine the confidence interval based on the upper and lower boundaries.

[0017] In some alternative embodiments, the device also includes a parameter update module for performing the following operations: If the current cycle node is the first cycle node, the running error is used as the mean running error, and the running error variance is configured based on the preset initial value; If the current cycle node is not the first cycle node, update the error statistics parameters using the running error, including the following steps: The mean of the operating error of the error statistics parameter is used as the old mean of the operating error, and the variance of the operating error of the error statistics parameter is used as the old variance of the operating error. The first coefficient is obtained by using the preset number of times as the numerator and the sum of the preset number of times and 1 as the denominator; where the preset number of times represents the number of times the error statistics parameter is updated. Using 1 as the numerator and the sum of the preset number and 1 as the denominator, the second coefficient is obtained by calculation; The new mean operating error is obtained by adding the product of the first coefficient and the old mean operating error to the product of the second coefficient and the operating error. Multiply the square of the old operating error variance by the first coefficient to obtain the first value; The second value is obtained by multiplying the square of the difference between the operating error and the mean of the new operating error by the second coefficient. The square root of the sum of the first and second values ​​is taken to obtain the new operating error variance; Update the error statistics parameters using the new mean and variance of the operating error.

[0018] To achieve the above objectives, another aspect of the present invention provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the aforementioned method.

[0019] To achieve the above objectives, another aspect of the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the aforementioned method.

[0020] To achieve the above objectives, another aspect of the present invention provides a computer program product, including a computer program that, when executed by a processor, implements the aforementioned method.

[0021] This invention provides at least the following beneficial effects: It offers a graded fault-tolerant monitoring method, device, electronic device, storage medium, and program product for a motor temperature sensor. This solution periodically acquires motor temperature and operating data. The temperature data includes the main sensing temperature collected by the main temperature sensor and the auxiliary monitoring temperature collected by auxiliary temperature monitoring points. The operating data includes operating parameters and operating errors. The auxiliary monitoring temperature and operating parameters are input into a pre-trained data-driven model to predict the predicted temperature. The prediction error is quantized based on the main sensing temperature and the predicted temperature. The prediction error is compared with a preset confidence interval. If the prediction error falls within the confidence interval, the main temperature sensor is determined to be reliable; otherwise, the main temperature sensor is determined to be unreliable. The system employs a confidence interval method. The confidence interval is determined based on the mean and variance of prediction errors obtained from historical prediction errors at several historical periodic nodes when the motor is in a normal state. When the main temperature sensor is in a reliable state, if the main sensor temperature does not exceed a preset high-temperature safety threshold, the motor's operating state is determined to be normal; otherwise, the motor's operating state is determined to be faulty. When the main temperature sensor is in an unreliable state, a joint fault probability is quantified based on the predicted temperature, the preset high-temperature safety threshold, the operating error, and preset error statistics parameters. If the joint fault probability is less than a preset probability judgment threshold, the motor's operating state is determined to be warning; otherwise, the motor's operating state is determined to be faulty. Based on the operating state, a graded response is performed using a preset mapping relationship. This invention periodically acquires the main sensor temperature, auxiliary monitoring temperature, and operating data including operating parameters and operating errors. The auxiliary monitoring temperature and operating parameters are input into a pre-trained data-driven model to obtain the predicted temperature, which is then compared with the main sensor temperature to obtain the prediction error. The reliability of the main temperature sensor is determined based on the comparison between the prediction error and the confidence interval obtained from historical prediction error statistics, thereby achieving a reliability assessment of the main sensor temperature and avoiding misjudgments caused by sensor distortion. In a reliable state, the system compares the main sensor temperature with a preset high-temperature safety threshold to determine whether the system is functioning normally or experiencing a fault. In an unreliable state, the system quantifies the joint fault probability based on the predicted temperature, the preset high-temperature safety threshold, operating error, and error statistics parameters. This probability is then compared with a preset probability judgment threshold to determine whether a warning or fault has occurred, leading to a tiered response. This method can provide redundant temperature predictions using a data-driven model when the main temperature sensor malfunctions, and achieve tiered fault-tolerant control through joint probability, reducing false alarms and unplanned downtime, and improving the reliability and safety of temperature monitoring. Attached Figure Description

[0022] Figure 1 This is a schematic diagram of an implementation environment for the graded fault-tolerant monitoring method for motor temperature sensors provided in this embodiment of the invention; Figure 2 This is a flowchart illustrating the graded fault-tolerant monitoring method for motor temperature sensors provided in an embodiment of the present invention. Figure 3 This is a schematic diagram of the overall architecture principle of the graded fault-tolerant monitoring of motor temperature sensors provided in this embodiment of the invention; Figure 4 This is a schematic diagram illustrating a data example of temperature fault probability provided in an embodiment of the present invention; Figure 5 This is a schematic diagram illustrating a data example of the probability of normal operation provided in an embodiment of the present invention; Figure 6 This is a schematic diagram of the structure of the graded fault-tolerant monitoring device for motor temperature sensors provided in an embodiment of the present invention; Figure 7 This is a schematic diagram of the structure of the electronic device provided in an embodiment of the present invention. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention. In the following description, when referring to the accompanying drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with those of this invention; they are merely examples of apparatuses and methods consistent with some aspects of the embodiments of this invention.

[0024] It is understood that the terms "first," "second," etc., used in this invention may be used to describe various concepts, but unless specifically stated otherwise, these concepts are not limited by these terms. These terms are only used to distinguish one concept from another. For example, without departing from the scope of embodiments of this invention, first information may also be referred to as second information, and similarly, second information may also be referred to as first information. Depending on the context, the words "if" or "when" as used herein may be interpreted as "when," "in response to determination," or "in the event of a determination."

[0025] The terms “at least one,” “multiple,” “each,” “any,” etc., used in this invention, “at least one” includes one, two, or more than two; “multiple” includes two or more than two; “each” refers to each of the corresponding multiple; and “any” refers to any one of the multiple.

[0026] Unless otherwise defined, all technical and scientific terms used in this invention have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used in this invention is for descriptive purposes only and is not intended to limit the invention.

[0027] In related technologies, most existing technologies remain in a binary processing mode of "switching when a fault is detected," lacking refined identification of the continuous health status of sensors and failing to form a complete fault-tolerant control closed loop.

[0028] In view of this, this invention provides a graded fault-tolerant monitoring method and related equipment for motor temperature sensors. This method periodically acquires motor temperature data and operating data. The temperature data includes the main sensing temperature collected by the main temperature sensor and the auxiliary monitoring temperature collected by auxiliary temperature monitoring points. The operating data includes operating parameters and operating errors. The auxiliary monitoring temperature and operating parameters are input into a pre-trained data-driven model to predict the predicted temperature. The prediction error is obtained by quantizing the main sensing temperature and the predicted temperature. The prediction error is compared with a preset confidence interval. If the prediction error falls within the confidence interval, the main temperature sensor is determined to be in a reliable state; otherwise, the main temperature sensor is determined to be in an unreliable state. The interval is determined based on the mean and variance of prediction errors obtained from the statistical analysis of historical prediction errors corresponding to several historical periodic nodes when the motor is in a normal state. When the main temperature sensor is in a reliable state, if the main sensor temperature does not exceed the preset high-temperature safety threshold, the motor's operating state is determined to be normal; otherwise, the motor's operating state is determined to be faulty. When the main temperature sensor is in an unreliable state, a joint fault probability is obtained by quantifying the predicted temperature, the preset high-temperature safety threshold, the operating error, and preset error statistical parameters. If the joint fault probability is less than the preset probability judgment threshold, the motor's operating state is determined to be warning; otherwise, the motor's operating state is determined to be faulty. Based on the operating state, a graded response is performed using a preset mapping relationship. This invention periodically acquires the main sensor temperature, the auxiliary monitoring temperature, and operating data including operating parameters and operating errors. The auxiliary monitoring temperature and operating parameters are input into a pre-trained data-driven model to obtain the predicted temperature, which is then compared with the main sensor temperature to obtain the prediction error. The reliable state of the main temperature sensor is determined based on the comparison between the prediction error and the confidence interval obtained from the statistical analysis of historical prediction errors, thereby achieving a reliability assessment of the main sensor temperature and avoiding misjudgment caused by sensor distortion. In a reliable state, the system compares the main sensor temperature with a preset high-temperature safety threshold to determine whether the system is functioning normally or experiencing a fault. In an unreliable state, the system quantifies the joint fault probability based on the predicted temperature, the preset high-temperature safety threshold, operating error, and error statistics parameters. This probability is then compared with a preset probability judgment threshold to determine whether a warning or fault has occurred, leading to a tiered response. This method can provide redundant temperature predictions using a data-driven model when the main temperature sensor malfunctions, and achieve tiered fault-tolerant control through joint probability, reducing false alarms and unplanned downtime, and improving the reliability and safety of temperature monitoring.

[0029] It is understood that the graded fault-tolerant monitoring method for motor temperature sensors provided by this invention can be applied to any computer device with data processing and computing capabilities, and this computer device can be various terminals or servers. When the computer device in the embodiment is a server, the server is an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN (Content Delivery Network), and big data and artificial intelligence platforms. Optionally, the terminal can be a smartphone, tablet, laptop, or desktop computer, but it is not limited to these.

[0030] like Figure 1 The diagram shown is a schematic representation of an implementation environment provided by an embodiment of the present invention. (Refer to...) Figure 1 The implementation environment includes at least one terminal 102 and a server 101. The terminal 102 and the server 101 can be connected via a network, either wirelessly or via a wired connection, to complete data transmission and exchange.

[0031] Server 101 can be a standalone physical server, a server cluster or distributed system consisting of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN (Content Delivery Network), and big data and artificial intelligence platforms.

[0032] Additionally, server 101 can also be a node server in a blockchain network. Blockchain is a novel application model of computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanisms, and encryption algorithms.

[0033] Terminal 102 can be a smartphone, tablet computer, laptop computer, desktop computer, smart speaker, smartwatch, etc., but is not limited to these. Terminal 102 and server 101 can be directly or indirectly connected via wired or wireless communication, and this embodiment of the invention does not impose any limitations.

[0034] For example, based on Figure 1 The implementation environment shown in this embodiment of the invention provides a graded fault-tolerant monitoring method for a motor temperature sensor. The following description uses the application of this graded fault-tolerant monitoring method for a motor temperature sensor in server 101 as an example. It can be understood that this graded fault-tolerant monitoring method for a motor temperature sensor can also be applied to terminal 102.

[0035] Reference Figure 2 , Figure 2 This is an optional flowchart of the graded fault-tolerant monitoring method for motor temperature sensors provided in the embodiments of the present invention. The executing entity of the graded fault-tolerant monitoring method for motor temperature sensors can be any of the aforementioned computer devices (including servers or terminals). Figure 2 The method may include, but is not limited to, steps S100 to S600.

[0036] Step S100: Periodically acquire motor temperature data and operating data; The temperature data includes the main sensing temperature collected by the main temperature sensor and the auxiliary monitoring temperature collected by the auxiliary temperature monitoring point; the operating data includes operating parameters and operating errors. It should be noted that, in some embodiments, acquiring the motor's temperature data and operating parameters may include the following steps: acquiring the main sensing temperature through a main temperature sensor installed at a preset position on the motor windings; wherein, the main temperature sensor includes a negative temperature coefficient thermistor or a platinum resistance temperature sensor; acquiring at least one of the following as auxiliary monitoring temperatures through temperature sensors preset on the motor's power module substrate, housing, and driver: the power module substrate temperature, the motor housing temperature, and the driver's ambient temperature; acquiring the motor's speed, power, and phase current as operating parameters; acquiring the command operation data and actual operation data of the joint module corresponding to the motor, and determining the operating error based on the difference between the command operation data and the actual operation data; wherein, the command operation data includes command values ​​for the position, speed, and current of the joint module, and the actual operation data includes the actual values ​​of the position, speed, and current of the joint module.

[0037] Specifically, based on a preset cycle, the above operation is performed once at each cycle node, which allows the motor's temperature data and operating data to be acquired periodically.

[0038] For example, in some specific embodiments, the main temperature sensor is installed at a predetermined location on the motor windings, such as the motor stator windings or a location near the windings. The main temperature sensor can be a negative temperature coefficient thermistor or a platinum resistance temperature sensor, such as Pt100 or Pt1000. The main temperature sensor acquires the main sensing temperature. .

[0039] The auxiliary temperature monitoring points include temperature sensors preset on the power module board, housing, and driver of the motor, which collect at least one of the following as auxiliary monitoring temperatures: power module board temperature, motor housing temperature, and driver ambient temperature.

[0040] The operating parameter acquisition module collects the motor's speed, power, and phase current as operating parameters.

[0041] The operational error is obtained by collecting both commanded and actual operational data from the joint module corresponding to the motor, and determining the operational error based on the difference between the commanded and actual operational data. Specifically, the commanded operational data includes the commanded values ​​for the position, speed, and current of the joint module, while the actual operational data includes the actual values ​​for the position, speed, and current of the joint module. ,in It represents operational data, including position, speed, and current; Indicates the error in the running data; This represents the instruction data given to the motor for operation; This represents the actual operational data read from the joint module.

[0042] Step S200: Input the auxiliary monitoring temperature and operating parameters into the pre-trained data-driven model to predict the predicted temperature, and quantify the prediction error based on the main sensor temperature and the predicted temperature. For example, in some specific implementations, auxiliary monitoring temperature and operating parameters are input into a pre-trained data-driven model, and the data-driven model outputs a predicted temperature for the current moment. Based on the main sensor temperature and predicted temperature Calculate prediction error .

[0043] It should be noted that, in some embodiments, the method may further include the following steps: acquiring training data; wherein, each training sample of the training data includes the historical main sensor temperature, historical auxiliary monitoring temperature and historical operating parameters of the same historical cycle node when the motor is in normal state, and different training samples of the training data correspond to different historical cycle nodes; Using historical auxiliary monitoring temperatures and historical operating parameters as input data, and historical main sensor temperatures as label data, a preset model is trained for temperature prediction. In response to the temperature prediction training results, the preset model is optimized and its parameters are tuned to obtain a trained model. The trained model is then validated using preset validation data to obtain validation results. If the accuracy of the validation results meets the preset accuracy conditions, the trained model is used as the data-driven model; otherwise, the trained model is used as the preset model, and the process returns to the steps of using historical auxiliary monitoring temperatures and historical operating parameters as input data, until the accuracy of the validation results meets the preset accuracy conditions. The preset model includes a neural network model, a support vector regression model, or an ensemble learning model.

[0044] In some optional implementations, the input to the data-driven model may also include a sequence of measured temperatures consisting of the main sensor temperatures corresponding to the trusted state at a preset time period prior to the current moment. Correspondingly, during the training process of the preset model, each training sample also includes a sequence of historical measured temperatures (consisting of the historical main sensor temperatures corresponding to each cycle node within the preset time period prior to the historical cycle node) at a trusted state at a preset time period prior to the historical cycle node.

[0045] For example, in some specific implementations, a pre-trained data-driven model (which may employ neural networks, support vector regression, or ensemble learning methods) takes as input auxiliary temperature monitoring point data, operating parameters (speed, power, current, etc.), and the last normal temperature data before the sensor becomes "unreliable," and outputs the predicted temperature value at the main temperature measurement location of the motor at the current moment.

[0046] Specifically, the model training process is as follows: First, training data is acquired. Each training sample includes the historical primary sensor temperature, historical auxiliary monitoring temperature, and historical operating parameters at the same historical cycle node when the motor is in normal condition. Different training samples correspond to different historical cycle nodes. Then, the historical auxiliary monitoring temperature and the historical operating parameters are used as input features, and the historical primary sensor temperature is used as a label to train the preset model for temperature prediction. The preset model can be a neural network model (such as a BP neural network, LSTM, etc.), a support vector regression (SVR) model, or an ensemble learning model (such as a random forest, XGBoost, etc.). After training, in response to the training results, the preset model is optimized and its parameters are tuned (such as adjusting the number of neural network layers, learning rate, regularization coefficient, or the kernel function parameters and penalty coefficient of the SVR, etc.) to obtain the trained model.

[0047] Next, the trained model is validated using pre-set validation data (i.e., an independent validation set, which can be pre-divided from the training data or collected separately), and the validation accuracy (such as mean squared error (MSE), mean absolute percentage error (MAPE), etc.) is calculated. If the accuracy corresponding to the validation result meets the preset accuracy conditions (e.g., MAPE is less than a preset threshold, or the coefficient of determination R² is greater than a preset value), then the current trained model is determined as the final data-driven model. If the validation accuracy does not meet the preset accuracy conditions, then the current trained model is used as the preset model, and the training and parameter tuning steps are re-executed (i.e., returning to the step of using historical auxiliary monitoring temperature and historical operating parameters as input data), iteratively optimizing until the validation accuracy meets the requirements.

[0048] After training, the model is deployed to the control unit of the joint module. When the main temperature sensor is in a reliable state, the control unit can periodically use the measured main sensor temperature, auxiliary monitoring temperature, and operating parameters to fine-tune the data-driven model to maintain model accuracy.

[0049] Specifically, when the input to the data-driven model also includes a sequence of measured temperatures consisting of the master sensor temperatures corresponding to a preset period of reliable conditions prior to the current moment, correspondingly, during the training process of the preset model, each training sample also includes a sequence of historical measured temperatures corresponding to a preset period of reliable conditions prior to the historical cycle node corresponding to that training sample. This historical measured temperature sequence consists of the historical master sensor temperatures corresponding to each cycle node within the preset period of time prior to the historical cycle node. For example, if the preset period is 10 minutes, then the measured temperature sequence is the sequence of master sensor temperatures for each control cycle within the 10 minutes prior to the current moment. This allows for the use of thermal inertia information to improve temperature prediction. The accuracy.

[0050] After the data-driven model is trained, it is deployed to the control unit of the joint module. When the main temperature sensor is in a reliable state, the control unit can periodically fine-tune the data-driven model using the measured main sensor temperature, auxiliary monitoring temperature, and operating parameters to maintain model accuracy.

[0051] Step S300: Compare the prediction error with a preset confidence interval. If the prediction error falls within the confidence interval, the main temperature sensor is determined to be in a reliable state; otherwise, the main temperature sensor is determined to be in an unreliable state. The confidence interval is determined based on the mean and variance of the prediction error obtained from the statistical analysis of historical prediction errors at several historical period nodes when the motor is in normal condition. It should be noted that, in some embodiments, the method may further include the following steps: multiplying the prediction error variance by a preset confidence coefficient to obtain the boundary redundancy; adding the boundary redundancy to the mean prediction error to obtain the upper boundary; subtracting the boundary redundancy from the mean prediction error to obtain the lower boundary; and determining the confidence interval based on the upper and lower boundaries.

[0052] For example, in some specific implementations, the prediction error is... The prediction error is compared with a preset confidence interval. The confidence interval is the mean prediction error obtained by statistically analyzing the historical prediction errors at several historical cycle nodes when the motor is in normal condition. and the standard deviation of prediction error Determine, for example, the confidence interval is [ ], To pre-set the confidence coefficient, it is recommended to take... =[2,3]. When If the temperature falls within the confidence interval, the main temperature sensor is determined to be in a reliable state; otherwise, the main temperature sensor is determined to be in an unreliable state.

[0053] Step S400: When the main temperature sensor is in a reliable state, if the main sensor temperature does not exceed the preset high temperature safety threshold, the motor's operating state is determined to be normal; otherwise, the motor's operating state is determined to be faulty. For example, in some specific embodiments, when the main temperature sensor is in a reliable state, the main sensor temperature is used as the reference temperature. As a monitoring basis. If the main sensor temperature The preset high temperature safety threshold has not been exceeded. If the main sensor temperature is normal, then the motor's operating status is determined to be normal; if the main sensor temperature is normal... Exceeding the preset high temperature safety threshold If so, the motor's operating state is determined to be a fault state.

[0054] It should be noted that the error statistics parameters include the mean of the operating error and the variance of the operating error. In some embodiments, when the main temperature sensor is in a reliable state, the method may further include the following steps: if the current cycle node is the first cycle node, use the operating error as the mean of the operating error, and configure the variance of the operating error based on a preset initial value; wherein the preset initial value is 0 or a very small number; if the current cycle node is not the first cycle node, update the error statistics parameters using the operating error, including the following steps: use the mean of the operating error of the error statistics parameters as the old mean of the operating error, and use the variance of the operating error of the error statistics parameters as the old variance of the operating error; use a preset number as the numerator, and set the preset... The sum of the number of iterations and 1 is used as the denominator to obtain the first coefficient; where the preset number of iterations represents the number of times the error statistics parameters are updated; 1 is used as the numerator, and the sum of the preset number of iterations and 1 is used as the denominator to obtain the second coefficient; the product of the first coefficient and the old operating error mean is added to the product of the second coefficient and the operating error to obtain the new operating error mean; the square of the old operating error variance is multiplied by the first coefficient to obtain the first value; the square of the difference between the operating error and the new operating error mean is multiplied by the second coefficient to obtain the second value; the square root of the sum of the first value and the second value is taken to obtain the new operating error variance; the error statistics parameters are updated using the new operating error mean and the new operating error variance.

[0055] For example, in some specific implementations, if the main sensor is in a reliable state, the main sensor data is used as the standard. If the temperature does not exceed a preset high-temperature safety threshold, the joint is in normal operation, and the operating error statistics parameters are updated according to the status of the executed command.

[0056]

[0057] in, This represents the mean of the new running error. This indicates that n updates have been performed. This represents the mean of the old operating error. Indicates operational error. This represents the variance of the new operating error. This represents the variance of the old operating error.

[0058] Step S500: When the main temperature sensor is in an unreliable state, the joint fault probability is obtained by quantification based on the predicted temperature, the preset high temperature safety threshold, the operating error and the preset error statistics parameters. If the joint fault probability is less than the preset probability judgment threshold, the motor's operating state is determined to be a warning state; otherwise, the motor's operating state is determined to be a fault state. It should be noted that the error statistics parameters include the mean and variance of the operating error. In some embodiments, the joint failure probability is quantified based on the predicted temperature, the preset high-temperature safety threshold, the operating error, and the preset error statistics parameters. This may include the following steps: subtracting the preset high-temperature safety threshold from the predicted temperature, and then adding the predicted error variance to obtain a third value; multiplying the third value by a preset confidence coefficient to obtain a fourth value; using the negative value of the fourth value as the exponent of the natural constant to obtain a fifth value; using the preset sixth value as the numerator and the sum of the fifth value and 1 as the denominator to calculate the temperature failure probability; and then multiplying 1 by the temperature... The difference between the degree of failure probability is taken as the normal temperature probability; the square of the difference between the operating error and the mean operating error is taken as the numerator, and twice the square of the variance of the operating error is taken as the denominator to obtain the seventh value; based on the negative value of the seventh value, the eighth value is obtained by calculating through the natural exponential function; the eighth value is multiplied by the preset ninth value to obtain the normal operating probability, and the difference between 1 and the normal operating probability is taken as the operating failure probability; the product of the operating failure probability and the temperature failure probability is taken as the numerator, and the product of the normal operating probability and the temperature failure probability plus the product of the operating failure probability and the temperature failure probability is taken as the denominator to obtain the joint failure probability.

[0059] For example, in some specific implementations, if the main sensor is in an unreliable state, the probability of failure is determined by combining the predicted temperature with the joint probability of operational errors:

[0060]

[0061]

[0062] in, Used to identify faults Used to indicate normal. Used to identify probability, Indicates the probability of a temperature-related failure. It is a natural constant. This indicates the preset confidence coefficient. Indicates the predicted temperature. This indicates the preset high-temperature safety threshold. This represents the variance of the prediction error. The sixth value (can be 0.9); To ensure normal operating probability, It is the ninth value. It is a natural exponential function. For operational error, This represents the average running error. This represents the variance of the operating error. Indicates the combined failure probability. Indicates the probability of normal temperature (through) get), Indicates the probability of operational failure (via get).

[0063] Step S600: Based on the operating status, perform hierarchical response using a preset mapping relationship; It should be noted that in some embodiments, step S600 may include the following steps: when the operating state is normal, maintain the normal operation of the motor; when the operating state is warning, maintain the normal operation of the motor and issue an abnormal reminder from the main temperature sensor; wherein, if the number of consecutive times the operating state is warning reaches a preset threshold, the motor is decelerated and stopped; when the operating state is faulty, the motor is decelerated and stopped.

[0064] For example, in some specific implementations, the hierarchical response can be implemented as follows: When the operating status is normal, the motor will continue to operate normally without issuing any alarms.

[0065] When the operating status is in the warning state, the motor continues to operate normally, and a main temperature sensor abnormality reminder is issued, prompting maintenance personnel to check or replace the main temperature sensor.

[0066] If the number of consecutive times the operating status is in a warning state reaches a preset threshold N, the motor will be slowly decelerated and stopped. The preset threshold N can be set according to the actual operating conditions, for example, N=20 sampling cycles. By setting a continuous preset threshold, misjudgments caused by instantaneous interference can be avoided, and deceleration and shutdown will only be performed when the main temperature sensor is continuously abnormal.

[0067] When the operating status is faulty, the motor is slowed down and stopped to protect the safety of the motor and joint module.

[0068] In some alternative implementations, when the main temperature sensor recovers from an unreliable state to a reliable state, it does not immediately revert to using the main sensor temperature as the monitoring basis, but instead enters an observation period. During the observation period, the predicted temperature continues to be used. As a monitoring basis, the main sensor temperature data is only used for comparison; simultaneously, the continuous reliable observation counter cnt_observe is started. If the prediction error occurs in each control cycle... If the data falls within the confidence interval, cnt_observe is incremented by 1. When cnt_observe is greater than or equal to the preset number N_observe (e.g., N_observe corresponds to the number of cycles between 1 and 2 minutes), the statistics are unfrozen and updated, reverting to using the main temperature sensor data as the monitoring basis, and incorporating the main sensor temperature data verified as reliable during the observation period. Update. If any control period during the observation period... If the value exceeds the confidence interval again, cnt_observe is cleared and the system re-enters the freeze period, while the continuous untrusted cycle counter cnt_untrust continues to increment. This effectively distinguishes between transient disturbances and persistent sensor failures, improving the stability of fault-tolerant control.

[0069] To explain in detail the principle of the technical solution of the present invention, the overall process of the present invention will be described below with reference to some specific embodiments. It is easy to understand that the following is an explanation of the technical principle of the present invention and should not be regarded as a limitation of the present invention.

[0070] In view of the shortcomings of the prior art, embodiments of the present invention provide a technical solution for graded fault-tolerant monitoring of motor temperature sensors. The technical solution of the present invention can be achieved through, as follows: Figure 3 The system architecture shown is specifically divided into four layers: perception layer, evaluation layer, prediction layer, and execution layer. The perception layer collects data from the main motor temperature sensor, auxiliary temperature measurement point data, and operating parameters; the evaluation layer performs a quantitative assessment of the main sensor's reliability based on a normal distribution; the prediction layer initiates temperature prediction through multi-source information fusion when the sensor is suspected; and the execution layer provides a graded response based on the evaluation results and prediction errors.

[0071] In some specific embodiments, the module functions of the system architecture of the technical solution of the present invention are described as follows: (1) Main temperature sensor of motor: Installed on or near the stator windings of the motor, this sensor, typically an NTC thermistor or platinum resistance thermometer (such as Pt100 / Pt1000), is responsible for real-time monitoring of the motor's core temperature. The output of this sensor is the core input signal of this system and the primary object of reliability assessment.

[0072] (2) Auxiliary temperature monitoring points: This includes, but is not limited to, NTC temperature sensors for the power module substrate, motor housing temperature sensors, and driver ambient temperature sensors. These sensors collect operating parameters such as the motor's current speed, output power, phase current, and load rate. These parameters have a clear physical relationship with temperature (e.g., copper loss is proportional to the square of the current) and can serve as input features for data-driven predictive models. These sensors are independent of the main temperature sensor and can serve as redundant information sources, providing auxiliary references in sensor evaluation and fusion prediction.

[0073] (3) Run the data acquisition module: The position error, speed error, and current error of the joint module are collected and calculated as follows:

[0074] It represents operational data, including position, speed, and current; Indicates the error in the running data; This represents the instruction data given to the motor for operation; This represents the actual operational data read from the joint module; (4) Sensor reliability assessment module: Based on historical readings from the main temperature sensor, temperature-related error statistics are established, and this data follows a normal distribution. This module uses the deviation of the current reading from the distribution parameters to determine the sensor's reliability. The specific judgment criteria are as follows: Confidence State: The difference between the measured motor temperature reading and the model-predicted temperature falls within [a certain range]. Within the interval ( To preset the confidence coefficient, such as taking =2 corresponds to the 95% confidence interval); Unreliable state: The difference between the measured motor temperature reading and the model-predicted temperature exceeds [a certain threshold]. interval; (5) Operational error statistics update module: When the sensor is in a "trusted state", the error statistics (mean) are updated after each sampling. and standard deviation This allows the module to track slow drift caused by long-term aging or environmental changes in sensors. When a sensor enters an "untrusted state," the module freezes statistical updates to prevent abnormal data from contaminating historical benchmarks.

[0075] (6) Multi-source information fusion prediction model: This is a pre-trained data-driven model (which can employ neural networks, support vector regression, or ensemble learning methods). The inputs are auxiliary temperature monitoring point data, operating parameters (speed, power, current, etc.), and the last normal temperature data before the sensor becomes "unreliable." The output is the predicted temperature value at the main temperature measurement location of the motor at the current moment. The model is trained during the equipment commissioning phase and fine-tuned periodically using measured data while the sensor is operating normally to maintain model accuracy.

[0076] (7) Error comparison and monitoring module: The system receives the sensor reliability assessment module's results. When the main motor sensor is in a reliable state, it performs temperature-based health monitoring and updates the operating error statistics parameters. When the main sensor is in a suspicious state, the system uses the predicted temperature and combines it with the joint probability of the operating error to determine the probability of damage.

[0077] (8) Hierarchical response module: Based on the credibility assessment and error monitoring results, a three-level response strategy is implemented: Normal: The motor sensor is functioning normally, and the motor temperature is within a safe range; Warning: The motor sensor is damaged, but the predicted temperature is within the safe range and the operating error is within the normal range; Fault: The motor sensor feedback temperature exceeds the preset high safety threshold or the motor predicted temperature exceeds the preset low safety threshold and the operating error exceeds the normal range; The system executes the following judgment loop in each control cycle: first, it assesses the reliability of the sensor, and then executes different processing paths according to the reliability status. When the sensor is normal, it takes the sensor as the main factor and updates the statistical benchmark. When it is suspicious, it starts the prediction model for monitoring. When the limit is exceeded, it performs a speed reduction and shutdown.

[0078] In some specific embodiments, the technical solution of the present invention can be implemented through the following process: Step S1: During each control cycle, three types of data are collected simultaneously: (a) the current reading of the motor main temperature sensor. (b) Data from auxiliary temperature monitoring points (power module NTC, casing temperature, etc.); (c) Operating parameters (speed ω, power P, phase current I, etc.).

[0079] Step S2: Calculate the predicted temperature based on the pre-trained model and the input running parameters. Calculate the prediction error. .

[0080] Step S3: Determine Whether it falls Within the range ( To pre-set the confidence coefficient, it is recommended to take... =[2,3]). If so, the main sensor is determined to be in a trusted state; otherwise, it is in an untrusted state. The mean error between the pre-collected predicted temperature and the measurement data from the main sensor during normal operation. The corresponding variance; Step S4: If the main sensor is in a reliable state, the main sensor data is used as the standard. If the temperature does not exceed the preset high-temperature safety threshold, the joint is in normal operation. The operating error statistics parameters are updated according to the status of the executed command.

[0081]

[0082] Among them, subscript For position, speed, or current, depending on the current operating mode (position mode, speed mode, or torque mode), subscript Used to indicate the state before an update, index Used to identify after a status update This represents the mean running error before the state update. This represents the mean running error after the state update. This represents the variance of the running error before the state update. This represents the variance of the runtime error after the state update, where n represents the number of update cycles executed.

[0083] If the temperature exceeds the preset high-temperature safety threshold (This threshold is determined by the motor insulation class and system thermal design), then it should be in a fault state.

[0084] Step S5: If the main sensor is in an unreliable state, determine the probability of damage based on the predicted temperature and the combined failure probability of the operating error:

[0085]

[0086]

[0087] The probability graph is as follows Figure 4 As shown (specifically, using a preset high-temperature safety threshold of 50℃ and a preset error standard deviation of 5℃ as an example of implementation scenario data, specifically, the inflection point of the peak of the correlation between predicted temperature and temperature failure probability is at predicted temperature = 45℃, and the intersection of the red dashed line and the data curve is the inflection point, at which point the corresponding temperature failure probability is 0.5). The probability graph is as follows Figure 5 As shown, with Increase to It is believed that the probability of a failure begins to increase at this point, until it increases to... They believed that the probability of a malfunction was at its highest. At the beginning, it is assumed that the system is operating normally with the highest probability. As the error increases, the probability of assuming the system is normal decreases. The maximum probability is never 1 in order to avoid masking the true nature of the system. and The information function.

[0088] Combined failure probability Less than the preset probability threshold Then it only enters the warning state (the motor is actually running normally), combined with the probability of failure. Greater than or equal to the preset probability judgment threshold Then it enters a fault state.

[0089] Step S6: Graded early warning module processing: In normal state, it runs normally and only updates the running error statistics; in warning state, it runs normally but reminds the main sensor of abnormality; in warning state, if it enters the warning state for N consecutive sampling cycles, it will slowly reduce speed and stop.

[0090] To distinguish between "transient interference misjudgment" and "persistent sensor failure," the system does not immediately unfreeze statistical updates when the sensor recovers from an "untrustworthy" state to a "trustworthy" state; instead, it enters an observation period. The specific logic is shown in Table 1 below: Table 1

[0091] In summary, this invention establishes a quantitative evaluation system for sensor reliability based on normal distribution, and constructs confidence intervals using historical error statistics. It replaces traditional empirical threshold judgment with statistical methods, and the evaluation results have a strict mathematical basis and interpretability. At the same time, the statistical benchmark is only updated when the sensor is reliable and automatically frozen when it is unreliable, which fundamentally prevents abnormal data from polluting the historical benchmark, and can adaptively track sensor aging and environmental drift.

[0092] In terms of fault tolerance, this invention constructs a multi-source information fusion redundancy architecture. When the main sensor fails, it automatically switches to a prediction model based on auxiliary temperature measurement points and operating parameters, achieving seamless fault tolerance. The prediction model uses methods such as neural networks, SVR, or ensemble learning, taking into input parameters that are physically related to temperature, such as rotational speed, power, and current, and outputting reliable temperature prediction values.

[0093] In terms of decision-making strategy, this invention introduces a joint probability judgment mechanism, which integrates temperature prediction and operational error to make probabilistic decisions, avoiding the "one-size-fits-all" drawbacks of hard threshold judgment. At the same time, it adopts a three-level progressive response strategy: continuous operation and baseline updates during normal operation, reminders for maintenance and N-cycle buffer during warnings, and speed reduction and shutdown during faults, significantly reducing false alarms and unplanned downtime losses.

[0094] like Figure 6 As shown, this embodiment of the invention also provides a graded fault-tolerant monitoring device 900 for a motor temperature sensor, which can implement the above-described method. This device may include: The data acquisition module 901 is used to periodically acquire the temperature data and operating data of the motor; the temperature data includes the main sensing temperature collected by the main temperature sensor and the auxiliary monitoring temperature collected by the auxiliary temperature monitoring point, and the operating data includes operating parameters and operating errors. Temperature prediction module 902 is used to input auxiliary monitoring temperature and operating parameters into a pre-trained data-driven model to predict the predicted temperature, and to obtain the prediction error based on the main sensor temperature and the predicted temperature. The sensing determination module 903 is used to compare the prediction error with the preset confidence interval. When the prediction error falls within the confidence interval, the main temperature sensor is determined to be in a reliable state; otherwise, the main temperature sensor is determined to be in an unreliable state. The confidence interval is determined based on the mean and variance of the prediction error obtained from the statistical analysis of the historical prediction errors corresponding to several historical cycle nodes when the motor is in a normal state. The temperature determination module 904 is used to determine the motor's operating status as normal if the main temperature sensor is in a reliable state and the main sensor temperature does not exceed the preset high temperature safety threshold; otherwise, it determines the motor's operating status as faulty. The probability determination module 905 is used to quantify the joint fault probability based on the predicted temperature, the preset high temperature safety threshold, the operating error and the preset error statistics parameters when the main temperature sensor is in an unreliable state. If the joint fault probability is less than the preset probability determination threshold, the motor's operating state is determined to be a warning state; otherwise, the motor's operating state is determined to be a fault state. The hierarchical response module 906 is used to perform hierarchical responses based on the operating status and using a preset mapping relationship.

[0095] In some alternative embodiments, the apparatus also includes a model training module for performing the following operations: Acquire training data; where each training sample of the training data includes the historical main sensor temperature, historical auxiliary monitoring temperature and historical operating parameters of the same historical cycle node when the motor is in normal state, and different training samples of the training data correspond to different historical cycle nodes. Historical auxiliary monitoring temperatures and historical operating parameters are used as input data, and historical main sensor temperatures are used as label data to train the preset model for temperature prediction. In response to the results of temperature prediction training, the preset model is optimized and its parameters are tuned to obtain the trained model. The training model is validated using preset validation data to obtain validation results. If the accuracy of the validation results meets the preset accuracy conditions, the training model is used as the data-driven model. Otherwise, the training model is used as the preset model, and the process returns to the step of using historical auxiliary monitoring temperature and historical operating parameters as input data until the accuracy of the validation results meets the preset accuracy conditions. The preset models include neural network models, support vector regression models, or ensemble learning models.

[0096] In some alternative embodiments, the device also includes an interval statistics module for performing the following operations: Multiply the prediction error variance by a preset confidence coefficient to obtain the boundary redundancy. The upper boundary is obtained by adding the mean of the prediction error to the boundary redundancy. Subtract the boundary redundancy from the mean prediction error to obtain the lower boundary; Determine the confidence interval based on the upper and lower boundaries.

[0097] In some alternative embodiments, the device also includes a parameter update module for performing the following operations: If the current cycle node is the first cycle node, the running error is used as the mean running error, and the running error variance is configured based on the preset initial value; If the current cycle node is not the first cycle node, update the error statistics parameters using the running error, including the following steps: The mean of the operating error of the error statistics parameter is used as the old mean of the operating error, and the variance of the operating error of the error statistics parameter is used as the old variance of the operating error. The first coefficient is obtained by using the preset number of times as the numerator and the sum of the preset number of times and 1 as the denominator; where the preset number of times represents the number of times the error statistics parameter is updated. Using 1 as the numerator and the sum of the preset number and 1 as the denominator, the second coefficient is obtained by calculation; The new mean operating error is obtained by adding the product of the first coefficient and the old mean operating error to the product of the second coefficient and the operating error. Multiply the square of the old operating error variance by the first coefficient to obtain the first value; The second value is obtained by multiplying the square of the difference between the operating error and the mean of the new operating error by the second coefficient. The square root of the sum of the first and second values ​​is taken to obtain the new operating error variance; Update the error statistics parameters using the new mean and variance of the operating error.

[0098] It is understood that the content of the above method embodiments is applicable to the present device embodiments. The specific functions implemented by the present device embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0099] This invention also provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the method described above. This electronic device can be any smart terminal, including tablet computers, in-vehicle computers, etc.

[0100] It is understood that the content of the above method embodiments is applicable to this device embodiment. The specific functions implemented by this device embodiment are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0101] like Figure 7 As shown, Figure 7 The hardware structure of an electronic device 1000 according to another embodiment is illustrated. The electronic device 1000 includes: The processor 1001 can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (aSIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of the present invention. The memory 1002 can be implemented as a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RaM). The memory 1002 can store the operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 1002 and is called and executed by the processor 1001. Input / output interface 1003 is used to implement information input and output; The communication interface 1004 is used to enable communication and interaction between this device and other devices. Communication can be achieved through wired means (such as USB, network cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.). Bus 1005 transmits information between various components of the device (e.g., processor 1001, memory 1002, input / output interface 1003, and communication interface 1004); The processor 1001, memory 1002, input / output interface 1003 and communication interface 1004 are connected to each other within the device via bus 1005.

[0102] The electronic device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0103] This invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method.

[0104] It is understood that the content of the above method embodiments is applicable to this storage medium embodiment. The specific functions implemented in this storage medium embodiment are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those achieved in the above method embodiments.

[0105] This invention also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0106] It is understood that the content of the above method embodiments is applicable to the embodiments of this program product. The specific functions implemented by the embodiments of this program product are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0107] Memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs and non-transitory computer-executable programs. Furthermore, memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, memory may optionally include memory remotely located relative to the processor, and these remote memories can be connected to the processor via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0108] The present invention provides a graded fault-tolerant monitoring method, device, electronic device, storage medium, and program product for motor temperature sensors. This method periodically acquires motor temperature and operational data. The temperature data includes the main sensing temperature collected by the main temperature sensor and the auxiliary monitoring temperature collected by auxiliary temperature monitoring points. The operational data includes operational parameters and operational errors. The auxiliary monitoring temperatures and operational parameters are input into a pre-trained data-driven model to predict the predicted temperature. The prediction error is quantized based on the main sensing temperature and the predicted temperature. The prediction error is compared with a preset confidence interval. If the prediction error falls within the confidence interval, the main temperature sensor is determined to be reliable; otherwise, it is determined to be unreliable. The confidence interval is determined based on the mean and variance of prediction errors obtained from the statistical analysis of historical prediction errors at several historical periodic nodes when the motor is in a normal state. When the main temperature sensor is in a reliable state, if the main sensor temperature does not exceed the preset high-temperature safety threshold, the motor's operating state is determined to be normal; otherwise, the motor's operating state is determined to be faulty. When the main temperature sensor is in an unreliable state, a joint fault probability is obtained based on the predicted temperature, the preset high-temperature safety threshold, the operating error, and preset error statistical parameters. If the joint fault probability is less than the preset probability judgment threshold, the motor's operating state is determined to be warning; otherwise, the motor's operating state is determined to be faulty. Based on the operating state, a graded response is performed using a preset mapping relationship. This invention periodically acquires the main sensor temperature, the auxiliary monitoring temperature, and operating data including operating parameters and operating errors. The auxiliary monitoring temperature and operating parameters are input into a pre-trained data-driven model to obtain the predicted temperature, which is then compared with the main sensor temperature to obtain the prediction error. The reliability of the main temperature sensor is determined based on the comparison between the prediction error and the confidence interval obtained from the statistical analysis of historical prediction errors, thereby achieving a reliability assessment of the main sensor temperature and avoiding misjudgments caused by sensor distortion. In a reliable state, the system compares the main sensor temperature with a preset high-temperature safety threshold to determine whether the system is functioning normally or experiencing a fault. In an unreliable state, the system quantifies the joint fault probability based on the predicted temperature, the preset high-temperature safety threshold, operating error, and error statistics parameters. This probability is then compared with a preset probability judgment threshold to determine whether a warning or fault has occurred, leading to a tiered response. This method can provide redundant temperature predictions using a data-driven model when the main temperature sensor malfunctions, and achieve tiered fault-tolerant control through joint probability, reducing false alarms and unplanned downtime, and improving the reliability and safety of temperature monitoring.

[0109] The embodiments described in this invention are for the purpose of more clearly illustrating the technical solutions of the embodiments of this invention, and do not constitute a limitation on the technical solutions provided by the embodiments of this invention. As those skilled in the art will know, with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by the embodiments of this invention are also applicable to similar technical problems.

[0110] Those skilled in the art will understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of the present invention, and may include more or fewer steps than shown, or combine certain steps, or different steps.

[0111] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0112] Those skilled in the art will understand that all or some of the steps in the methods disclosed above, as well as the functional modules / units in the systems and devices, can be implemented as software, firmware, hardware, or suitable combinations thereof.

[0113] The preferred embodiments of the present invention have been described above with reference to the accompanying drawings, but this does not limit the scope of the claims of the present invention. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and spirit of the present invention should be within the scope of the claims of the present invention.

Claims

1. A motor temperature sensor hierarchical fault-tolerant monitoring method, characterized by, The method includes the following steps: The temperature data and operating data of the motor are acquired periodically; wherein, the temperature data includes the main sensing temperature collected by the main temperature sensor and the auxiliary monitoring temperature collected by the auxiliary temperature monitoring point, and the operating data includes operating parameters and operating errors; The auxiliary monitoring temperature and the operating parameters are input into a pre-trained data-driven model to predict the temperature, and the prediction error is obtained by quantifying the main sensor temperature and the predicted temperature. The prediction error is compared with a preset confidence interval. If the prediction error falls within the confidence interval, the main temperature sensor is determined to be in a reliable state; otherwise, the main temperature sensor is determined to be in an unreliable state. The confidence interval is determined based on the mean and variance of the prediction error obtained by statistically analyzing the historical prediction errors corresponding to several historical cycle nodes when the motor is in a normal state. When the main temperature sensor is in the trusted state, if the main sensor temperature does not exceed the preset high temperature safety threshold, the motor is determined to be in a normal operating state; otherwise, the motor is determined to be in a fault operating state. When the main temperature sensor is in the unreliable state, the joint fault probability is quantified based on the predicted temperature, the preset high temperature safety threshold, the operating error and the preset error statistics parameters. If the joint fault probability is less than the preset probability judgment threshold, the operating state of the motor is determined to be the warning state; otherwise, the operating state of the motor is determined to be the fault state. Based on the operating status, a hierarchical response is performed using a preset mapping relationship; The error statistics parameters include the mean operating error and the variance of the operating error. When the main temperature sensor is in the reliable state, the method further includes the following steps: If the current cycle node is the first cycle node, the running error is used as the mean of the running error, and the running error variance is configured based on a preset initial value; wherein the preset initial value is 0 or a very small number; If the current cycle node is not the first cycle node, the error statistics parameters are updated using the running error, including the following steps: The mean of the operating error of the error statistics parameter is used as the old mean of the operating error, and the variance of the operating error of the error statistics parameter is used as the old variance of the operating error. The first coefficient is obtained by calculating the preset number of times as the numerator and the sum of the preset number of times and 1 as the denominator; wherein, the preset number of times represents the number of times the error statistics parameter is updated; Using 1 as the numerator and the sum of the preset number and 1 as the denominator, the second coefficient is obtained by calculation; Add the product of the first coefficient and the old operating error mean to the product of the second coefficient and the operating error to obtain the new operating error mean; Multiply the square of the old operating error variance by the first coefficient to obtain the first value; The second value is obtained by multiplying the square of the difference between the operating error and the mean of the new operating error by the second coefficient; The square root of the sum of the first value and the second value is taken to obtain the new operating error variance; The error statistics parameters are updated using the new mean and the new variance of the operating error. The error statistics parameters include the mean and variance of the operating error. The method for quantifying the joint failure probability based on the predicted temperature, the preset high-temperature safety threshold, the operating error, and the preset error statistics parameters includes the following steps: Subtract the preset high-temperature safety threshold from the predicted temperature, and then add the prediction error variance to obtain the third value; Multiply the third value by the preset confidence coefficient to obtain the fourth value; The negative value of the fourth value is used as the exponent of the natural constant to obtain the fifth value; The temperature failure probability is calculated by taking the preset sixth value as the numerator and the sum of the fifth value and 1 as the denominator. The difference between 1 and the temperature failure probability is taken as the normal temperature probability. The seventh value is obtained by taking the square of the difference between the operating error and the mean of the operating error as the numerator and twice the square of the variance of the operating error as the denominator. Based on the negative value of the seventh value, the eighth value is obtained by operating on the natural exponential function; Multiply the eighth value by the preset ninth value to obtain the probability of normal operation, and take the difference between 1 and the probability of normal operation as the probability of operation failure. The product of the operational failure probability and the temperature failure probability is used as the numerator, and the product of the normal operation probability and the temperature normal operation probability plus the product of the operational failure probability and the temperature failure probability is used as the denominator to calculate the joint failure probability.

2. The method of claim 1, wherein, The process of acquiring the motor's temperature data and operating parameters includes the following steps: The main sensing temperature is acquired by the main temperature sensor installed at a preset position on the winding of the motor; wherein, the main temperature sensor includes a negative temperature coefficient thermistor or a platinum resistance temperature sensor. The auxiliary monitoring temperature is obtained by collecting at least one of the following: the power module board temperature, the motor housing temperature, and the driver ambient temperature, using temperature sensors preset on the power module board, housing, and driver of the motor. The motor's speed, power, and phase current are collected as the operating parameters; Collect command operation data and actual operation data of the joint module corresponding to the motor, and determine the operation error based on the difference between the command operation data and the actual operation data; The command execution data includes command values ​​for the position, speed, and current of the joint module, while the actual execution data includes the actual values ​​for the position, speed, and current of the joint module.

3. The method of claim 1, wherein, The method further includes the following steps: Acquire training data; wherein, each training sample of the training data includes the historical main sensor temperature, historical auxiliary monitoring temperature and historical operating parameters of the same historical cycle node when the motor is in normal state, and different training samples of the training data correspond to different historical cycle nodes; The historical auxiliary monitoring temperature and the historical operating parameters are used as input data, and the historical main sensor temperature is used as tag data to train the preset model for temperature prediction. In response to the results of the temperature prediction training, the preset model is optimized and its parameters are tuned to obtain the trained model; The training model is validated using preset validation data to obtain validation results. If the accuracy corresponding to the validation results meets the preset accuracy conditions, the training model is used as the data-driven model. Otherwise, the training model is used as the preset model, and the process returns to the step of using the historical auxiliary monitoring temperature and the historical operating parameters as input data until the accuracy corresponding to the validation results meets the preset accuracy conditions. The preset model includes a neural network model, a support vector regression model, or an ensemble learning model.

4. The method according to claim 1, characterized in that, The method further includes the following steps: Multiply the prediction error variance by a preset confidence coefficient to obtain the boundary redundancy. The upper boundary is obtained by adding the mean of the prediction error to the boundary redundancy. Subtracting the boundary redundancy from the mean prediction error yields the lower boundary. The confidence interval is determined based on the upper and lower boundaries.

5. The method according to claim 1, characterized in that, The step of performing a tiered response based on the operating state using a preset mapping relationship includes the following steps: When the operating state is the normal state, maintain the normal operation of the motor; When the operating state is the warning state, maintain the normal operation of the motor and issue an abnormality reminder for the main temperature sensor; If the number of consecutive times the operating state is in the warning state reaches a preset threshold, the motor will be slowed down and stopped. When the operating state is a fault state, the motor is decelerated and stopped.

6. A graded fault-tolerant monitoring device for motor temperature sensors, characterized in that, The apparatus for implementing the method of claim 1 includes: The data acquisition module is used to periodically acquire the motor's temperature data and operating data; wherein, the temperature data includes the main sensing temperature collected by the main temperature sensor and the auxiliary monitoring temperature collected by the auxiliary temperature monitoring point, and the operating data includes operating parameters and operating errors; The temperature prediction module is used to input the auxiliary monitoring temperature and the operating parameters into a pre-trained data-driven model to predict the predicted temperature, and to quantify the prediction error based on the main sensor temperature and the predicted temperature. The sensing determination module is used to compare the prediction error with a preset confidence interval. When the prediction error falls within the confidence interval, the main temperature sensor is determined to be in a reliable state; otherwise, the main temperature sensor is determined to be in an unreliable state. The confidence interval is determined based on the mean and variance of the prediction error obtained by statistically analyzing the historical prediction errors corresponding to several historical cycle nodes when the motor is in a normal state. The temperature determination module is used to determine that the motor's operating state is normal if the main temperature sensor is in the trusted state and the main sensor temperature does not exceed a preset high temperature safety threshold; otherwise, it determines that the motor's operating state is faulty. The probability determination module is used to determine the joint fault probability based on the predicted temperature, the preset high temperature safety threshold, the operating error and the preset error statistics parameters when the main temperature sensor is in the unreliable state. If the joint fault probability is less than the preset probability determination threshold, the operating state of the motor is determined to be the warning state; otherwise, the operating state of the motor is determined to be the fault state. The hierarchical response module is used to perform hierarchical responses based on the operating state and using a preset mapping relationship.

7. An electronic device, characterized in that, The electronic device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the method according to any one of claims 1 to 5.

8. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the method described in any one of claims 1 to 5.

Citation Information

Patent Citations

  • Battery pack temperature sensor failure detection and fault tolerance method and system and vehicle

    CN121702577A

  • Coal spontaneous combustion temperature prediction method based on multi-source heterogeneous data adaptive fusion

    CN122288023A