Brushless direct current motor thermal management control optimization method based on multi-sensor network

Abstract

The application relates to the technical field of data processing, in particular to a brushless direct-current motor thermal management control optimization method based on a multi-sensor network. The method comprises the following steps: based on the abnormal characteristic correlation of temperature data at each position of the motor with stator phase current and motor rotating speed, correlation characteristic factors of each kind of temperature data are constructed; combined with abnormal scores of various temperature data at each time, first characteristic coefficients at each time are constructed; combined with the correlation between each kind of temperature data and all other kinds of temperature data and the similarity between the first characteristic coefficients at all times and temperature data abnormal characteristics, comprehensive characteristic values of various temperature data are constructed; attenuation parameters during temperature prediction of various temperatures are determined; combined with a prediction algorithm, various temperatures are predicted to perform motor thermal management control. The temperature control in the thermal management process of the brushless direct-current motor is improved, accurate motor thermal management processing is performed, and the stability of the brushless direct-current motor operation is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, in particular to a brushless direct current motor thermal management control optimization method based on a multi-sensor network. BACKGROUND

[0002] The brushless direct current motor (BLDC) has the advantages of high power density, high efficiency and low noise, and is widely used in new energy vehicle driving systems, industrial servo devices and aerospace actuators, etc. In the process of operation of the brushless direct current motor, the stator winding current loss, the core hysteresis and eddy current loss, and the bearing friction and windage loss will continuously generate heat, causing the temperature inside the motor to rise. If the temperature of the brushless direct current motor cannot be efficiently and accurately controlled, problems such as aging and failure of the stator winding insulation layer, demagnetization of the permanent magnet, and carbonization of the bearing grease will occur, affecting the stability and service life of the motor. Nowadays, a thermal management control optimization method based on a multi-sensor network is commonly used, which realizes accurate thermal state perception and dynamic adjustment of heat dissipation strategy by collecting the temperature of the key components of the motor and the associated operating parameters in real time, thereby improving the stability of the BLDC motor under high load and complex working conditions.

[0003] In the process of thermal management control of the BLDC motor based on the multi-sensor network, due to the complexity of the actual operating conditions, the analysis of the thermal state of the motor during the operation monitoring process deviates greatly, thereby affecting the accuracy of the thermal management processing of the BLDC motor. The existing thermal management control method for the BLDC motor through the multi-sensor network does not fully consider the influence of the load change rate and the environmental change on the heat dissipation demand of the motor, resulting in a mismatch between the heat dissipation performance under the motor thermal management control and the actual thermal state of the motor. For example, excessive heat dissipation under low load conditions will cause energy waste, and heat dissipation lag under high load conditions will cause temperature rise, etc. Therefore, the existing thermal management control of the brushless direct current motor has poor control effect, thereby reducing the stability of the motor operation. SUMMARY

[0004] In order to solve the above technical problems, the present application provides a brushless direct current motor thermal management control optimization method based on a multi-sensor network to solve the existing problems.

[0005] The brushless direct current motor thermal management control optimization method based on a multi-sensor network of the present application adopts the following technical scheme:

[0006] One embodiment of the present application provides a brushless direct current motor thermal management control optimization method based on a multi-sensor network, which comprises the following steps:

[0007] real-time collection of stator phase current, motor speed and temperature data at each position of the brushless direct current motor;

[0008] An abnormal score of each kind of data at each time is obtained by an abnormality detection algorithm to construct an abnormal score sequence of each kind of data; the stator phase current and the motor speed are taken as characteristic parameters; a motor thermal state abnormality associated characteristic factor of each kind of temperature data is constructed based on the difference between each kind of temperature data and the abnormal score sequence of all kinds of characteristic parameters, and the overall distribution size of all abnormal scores, and a first characteristic coefficient at each time is constructed by combining the abnormal scores of each kind of temperature data at each time;

[0009] Based on the first characteristic coefficient, time is divided, and in each time period, a temperature dynamic associated characteristic value of each kind of temperature data in each time period is constructed according to the correlation between each kind of temperature data and the time sequence of all other kinds of temperature data, and the similarity between the first characteristic coefficient at all times and all abnormal scores of each kind of temperature data;

[0010] Based on the temperature dynamic associated characteristic value and the associated characteristic factor, a comprehensive characteristic value of each kind of temperature data is constructed, the decay parameters of each kind of temperature prediction are determined, each kind of temperature is predicted by combining a prediction algorithm, and motor thermal management control is performed.

[0011] In one embodiment, the associated characteristic factor is obtained as follows:

[0012] The Euclidean distance between each characteristic parameter and the abnormal score sequence of each kind of temperature data is calculated; the fusion value of each characteristic parameter and the abnormal score of each kind of temperature data at all times is calculated; the abnormal association degree between each characteristic parameter and each kind of temperature data is positively correlated with the Euclidean distance and the fusion value, respectively;

[0013] The sum of the abnormal association degrees between each kind of temperature data and all kinds of characteristic parameters is taken as the associated characteristic factor of each kind of temperature data.

[0014] In one embodiment, the fusion value of the abnormal score is the sum of the mean value of all times of each characteristic parameter and the mean value of all times of each temperature data.

[0015] In one embodiment, the abnormal association degree is the product of the Euclidean distance and the fusion value.

[0016] In one embodiment, the first characteristic coefficient is obtained as follows:

[0017] The product of the associated characteristic factor of each kind of temperature data and the abnormal score of each kind of temperature data at each time is taken as a first product; the sum of the first products of all kinds of temperature data at each time is taken as the first characteristic coefficient at each time.

[0018] In one embodiment, the time division manner is:

[0019] The first feature coefficients of all time points are sorted in time sequence and curve fitting is performed, the slope of the corresponding data point in the fitting curve is calculated for each time point, the time point with the slope not equal to 0 and the first feature coefficient equal to 1 is taken as the division point for division, and each time period is obtained.

[0020] In one embodiment, the temperature dynamic correlation feature value is obtained by:

[0021] In each time period, the correlation degree average between each kind of temperature data and the time sequence of all other kinds of temperature data is calculated by a similarity algorithm, and is recorded as a first response value; the similarity between the set composed of the first feature coefficients of all time points in the time period and the set composed of the anomaly scores of each kind of temperature data at all time points is calculated, and is recorded as a second response value;

[0022] The temperature dynamic correlation feature value of each kind of temperature data in each time period is positively correlated with the second response value and negatively correlated with the first response value.

[0023] In one embodiment, the temperature dynamic correlation feature value is: the calculation result of the ratio of the second response value to the sum of the first response value and a preset adjustment parameter.

[0024] In one embodiment, the comprehensive feature value is the product of the mean value of the temperature dynamic correlation feature value of each kind of temperature data in all time periods and the correlation feature factor of each kind of temperature data.

[0025] In one embodiment, the decay parameter at the time of various temperature prediction is: the normalized value of the comprehensive feature value of various temperature data.

[0026] The present application has at least the following beneficial effects:

[0027] The application is directed to the fact that, in the thermal management control process of a brushless direct current motor, the influence of load change rate and environmental change on motor heat dissipation demand is not fully considered, resulting in a mismatch between the heat dissipation performance of the motor under thermal management control and the actual thermal state of the motor; a brushless direct current motor thermal management control optimization method based on a multi-sensor network is proposed, which first uses a multi-sensor network to adopt a hierarchical deployment and distributed collection method to collect state monitoring data in the motor thermal management process, further analyzes the temperature dynamic abnormal correlation characteristics of different positions of the motor under the influence of different dominant factors, based on the analysis results, more accurate temperature anomaly analysis of the motor is carried out, and the precise temperature deviation of the motor is determined, accurate signal feedback is carried out on the control under water cooling and cooling treatment, the temperature control in the thermal management process of the brushless direct current motor is improved, the precise motor thermal management treatment is implemented, and the stability of the brushless direct current motor operation is improved. BRIEF DESCRIPTION OF DRAWINGS

[0028] In order to more clearly illustrate the technical solutions and advantages of the embodiments of the present application or the prior art, the drawings needed in the description of the embodiments or the prior art will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and those skilled in the art can obtain other drawings according to these drawings without creative labor.

[0029] Figure 1 The flowchart of the brushless direct current motor thermal management control optimization method based on the multi-sensor network provided by the present application is shown in the figure.

[0030] Figure 2 The schematic diagram of the correlation characteristic factor acquisition process is shown in the figure. DETAILED DESCRIPTION

[0031] In order to further illustrate the technical means and effects adopted by the present application to achieve the predetermined invention purpose, the specific implementation, structure, features and effects of the brushless direct current motor thermal management control optimization method based on the multi-sensor network according to the present application are described in detail as follows. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, the specific features, structures or characteristics in one or more embodiments can be combined in any suitable form.

[0032] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the present application belongs.

[0033] The application provides a specific scheme of a brushless direct current motor thermal management control optimization method based on a multi-sensor network.

[0034] The application provides a specific scheme of a brushless direct current motor thermal management control optimization method based on a multi-sensor network.

[0035] The application provides a specific scheme of a brushless direct current motor thermal management control optimization method based on a multi-sensor network. Please refer to Figure 1 The method comprises the following steps.

[0036] In step S1, the stator phase current, the motor speed and the temperature data of each position of the brushless direct current motor are collected in real time.

[0037] In the application, the multi-sensor network is used to optimize the control of the thermal management of the brushless direct current motor. The multi-sensor network adopts a hierarchical deployment and distributed collection mode to collect and monitor the parameters in the operation process of the brushless direct current motor in real time. Specifically,

[0038] A K-type thermocouple sensor is arranged at the end of the stator winding to collect the winding temperature in real time. A patch-type platinum resistance is arranged outside the stator core to collect the core temperature in real time. A non-contact infrared temperature sensor is arranged outside the rotor permanent magnet to collect the surface temperature of the permanent magnet in real time. A micro temperature sensor is embedded inside the bearing end cover to collect the bearing temperature in real time. A Hall current sensor is connected in series on the cable connecting the motor controller and the stator winding to collect the stator phase current in real time. An optical speed sensor is installed at the end of the motor output shaft to collect the motor speed data in real time. Flow sensors are installed on the inlet and outlet pipelines of the cooling system to collect the flow data of the inlet and outlet pipelines of the cooling system in real time.

[0039] Further, the collected motor monitoring data is transmitted and preprocessed. Specifically, the collected current and speed periodic data are filtered by a mean filter at the edge node, and the temperature and flow slowly changing data are filtered and denoised by a Kalman filter, so as to reduce the influence of noise interference on the data quality in the data collection process of the brushless direct current motor. The preprocessed data is standardized, and the standardized data is transmitted to a thermal management control feature analysis module through a CAN bus for motor thermal state feature analysis. The transmission rate is 500 kbps to 1 Mbps, and the transmission period is 10 to 20 ms. The mean filter, the Kalman filter and the data standardization are all known contents, and the specific process will not be described here.

[0040] In step S2, the abnormal score of each kind of data at each time is obtained through an abnormality detection algorithm to construct an abnormal score sequence of each kind of data; the stator phase current and the motor speed are taken as characteristic parameters; the difference between each kind of temperature data and the abnormal score sequence of all kinds of characteristic parameters, and the overall distribution size of all abnormal scores, are used to construct the associated characteristic factor of the motor thermal state abnormality of each kind of temperature data, and the abnormal score of each kind of temperature data at each time is used to construct the first characteristic coefficient at each time.

[0041] In the process of analyzing the thermal state of the brushless direct current motor through the multi-sensor network, the dynamic change characteristics of the corresponding thermal state when the motor operating state changes should be fully considered, and then the trend of the thermal state change and the heat dissipation demand of the brushless direct current motor are accurately judged. During the operation of the brushless direct current motor, the key parameters such as the temperature of the core heating component, the stator phase current and the speed will change dynamically and be associated with the timing characteristics when the motor load fluctuates and the working condition switches, for example, in the condition of rapid acceleration, the current abnormally increases, which causes the increase of the copper loss of the motor winding and the abnormal increase of the winding temperature. During the change of the motor thermal state, the temperature change will be relatively lagged behind the change of the electrical parameters. Therefore, based on the above analysis, in order to accurately judge the change characteristics of the motor thermal state and more accurately analyze the heat dissipation demand of the motor, the dynamic change characteristics of the thermal state during the monitoring process should be analyzed according to the received monitoring data collected by the multi-sensor network, and the specific analysis and processing process is as follows:

[0042] During the operation of the brushless direct current motor, the difference of the motor operating state caused by the load change is different, that is, the dominant factor and the associated strength of the coupled change of the motor monitoring data are greatly different under different working conditions. Therefore, considering the influence of the deviation change of the operating state of the brushless direct current motor under the load change on the thermal state change, the stator phase current, the motor speed and the collected temperature data during the operation of the motor are analyzed.

[0043] For each kind of monitoring data of the motor, the LOF abnormality detection algorithm is used to process all the monitoring data at each time to obtain the LOF value of the monitoring data at each time, and all the LOF values are sorted in time sequence. The sorted data sequence is taken as the abnormal score sequence of the monitoring data during the operation of the motor. The LOF abnormality detection algorithm is a known technology, and the specific process is not described here.

[0044] It should be noted that for the abnormality detection of all collected data, this application only provides an abnormality detection method. There are many existing abnormality detection methods, and the implementer can also use other abnormality detection algorithms to detect the abnormality of all collected data to obtain the abnormal score of each kind of monitoring data at each time. This application does not make specific limitations.

[0045] Furthermore, considering the different dominant factors causing motor state deviations under load fluctuations, the dynamic correlation characteristics between motor parameter changes and temperature changes at different locations differ during motor state changes. Therefore, stator phase current and motor speed data are used as characteristic parameters of the state deviation response. The Euclidean distance between each characteristic parameter and the abnormal score sequence of temperature data at each location is calculated. The larger the Euclidean distance, the more significant the state anomaly between the motor thermal state change and the motor parameters. On the other hand, considering the difference between the abnormal response state of each characteristic parameter and the abnormal response of temperature, the mean of all elements in the abnormal score sequence of each characteristic parameter and each temperature data is calculated. The larger the mean, the more significant the overall abnormal response of each characteristic parameter or each temperature data in the motor state change. Based on the above analysis, the abnormal correlation degree of each characteristic parameter and each temperature data under different dominant factors is calculated. The formula for the calculation is:

[0046]

[0047] in, Indicates the first Characteristic parameters and the first The degree of abnormal correlation between different temperature data points and their abnormal operating status. Indicates the first The Euclidean distance between a feature parameter and the anomaly score sequence of the y-th temperature data; and They represent the first The mean of all elements in the anomaly score sequence of the y-th temperature data and the characteristic parameter.

[0048] The larger the value, the more significant the difference in abnormal response under different characteristic parameters as the dominant factors of state change; the larger the calculated abnormal correlation degree, the more significant the abnormal correlation change under different dominant factors caused by the change of motor state.

[0049] Furthermore, based on the above analysis, the anomalous characteristics of the thermal state under the influence of different dominant factors are calculated, specifically:

[0050] The sum of the abnormal correlations between each temperature data point of the motor and all types of characteristic parameters is calculated and used as a characteristic value of the abnormal response of the motor's thermal state under the influence of different dominant factors. This value is denoted as the correlation characteristic factor. The larger the correlation characteristic factor, the more significant the correlation between the corresponding temperature change and the abnormal thermal state of the motor. Furthermore, a comprehensive analysis is performed on the abnormal thermal state characteristics of the motor at different times during operation under the influence of different dominant factors, and the first characteristic coefficient of the abnormal thermal state of the motor is calculated. The calculation formula is as follows:

[0051]

[0052] wherein, represents the first characteristic coefficient of the thermal state anomaly of the motor under the influence of different dominant factors at the tth moment; represents the number of temperature data categories; represents the LOF value of the yth temperature data collected at the tth moment; represents the correlation characteristic factor of the yth temperature data on the thermal state anomaly of the motor; represents the Softmax function. The greater the first characteristic coefficient calculated is, the more significant the thermal state change anomaly under the influence of different dominant factors during the operation of the motor is.

[0053] Step S3, based on the first characteristic coefficient, dividing the time, in each time period, according to the correlation between each temperature data and the time sequence of all other categories of temperature data, and the similarity between the first characteristic coefficient of all moments and all abnormal scores of each temperature data, constructing the temperature dynamic correlation characteristic value of each temperature data in each time period.

[0054] Based on the above analysis, combined with the difference characteristics of the change of the motor operating state caused by the change of the load during the operation of the motor, the change characteristics of the thermal state of the motor under the influence of different dominant factors are accurately analyzed, and the accuracy of the operating state analysis in the thermal management process of the brushless direct current motor based on the multi-sensor network is improved. Specifically:

[0055] The first characteristic coefficient of each moment obtained by the above calculation is sorted in time sequence, and the least square method is used for curve fitting to calculate the slope of the corresponding data point in the fitting curve at each moment. The moment when the slope is not equal to 0 and the first characteristic coefficient of the corresponding moment is equal to 1 is divided as a characteristic moment to obtain each time period. The purpose of division is: if the LOF value is greater than 1, it indicates that the abnormal characteristic is significant, and the temperature data change interval is divided based on the moment of abnormal change, and then the temperature characteristics under different dominant factors in the abnormal change interval are effectively analyzed.

[0056] Furthermore, within each time period, data for all moments of each temperature data type are acquired and arranged in chronological order of acquisition time to obtain a time series sequence of each temperature data type; furthermore, within this time period, the Pearson correlation coefficient between any temperature data type and the time series sequences of each other temperature data type is calculated, and the mean of the absolute values ​​of all Pearson correlation coefficients for any temperature data type is taken as the first response value of the any temperature data type, which reflects the dynamic correlation characteristics of different locations under the influence of different dominant factors;

[0057] The Jaccard coefficient is calculated between the set of LOF values ​​for all time points corresponding to each temperature data type within each time period and the set of first characteristic coefficients for all time points within that time period. This Jaccard coefficient is used as the second response value, reflecting the characteristics of each temperature acquisition location's anomaly relative to the overall temperature under the influence of different dominant factors. Both the Pearson correlation coefficient and the Jaccard coefficient are well-known techniques, and their specific processes will not be elaborated upon.

[0058] It should be noted that, for the calculation of similarity between time series of temperature data, this application only provides one similarity calculation method. There are many existing similarity calculation methods, and implementers may also use other similarity algorithms to calculate the similarity between time series of temperature data. This application does not impose any specific restrictions.

[0059] Furthermore, based on the above analysis, taking any given time period as the current time period, we calculate the characteristic values ​​of the abnormal temperature dynamic correlation of the brushless DC motor under load fluctuations caused by different dominant factors. The calculation formula is as follows:

[0060]

[0061] in, Indicates the first time in the current time period The characteristic values ​​of abnormal changes in temperature dynamic correlation caused by different dominant factors are denoted as temperature dynamic correlation characteristic values. Indicates the first time in the current time period The second response value corresponding to the temperature data; Indicates the first time in the current time period The first response value corresponding to the temperature data; This represents a preset adjustment parameter, with a value range of (0, 0.1). This ensures that the denominator is not zero, and setting a small value has a negligible impact on the calculation result. In this embodiment, its value is set to 0.001. The larger the calculated feature value, the greater the likelihood that the temperature anomaly at the corresponding location is caused by abnormal changes in the thermal state of the motor, based on the analysis of temperature correlation anomalies at different locations under the influence of different dominant factors.

[0062] Step S4, based on the temperature dynamic correlation characteristic value and the correlation characteristic factor, constructing the comprehensive characteristic value of various temperature data, determining the attenuation parameter of various temperature prediction, combining the prediction algorithm, predicting various temperatures to perform motor thermal management control.

[0063] In order to accurately analyze the dynamic change characteristics of each position under load change in the process of motor thermal management, the mean value of the temperature dynamic correlation characteristic value corresponding to each temperature data of all time periods is calculated, and the product of the mean value and the correlation characteristic factor of each temperature data is taken as the characteristic value of the abnormal change of each temperature data due to the influence of different dominant factors on the change of motor thermal state during the operation of brushless motor. It is recorded as the comprehensive characteristic value of each temperature data. The greater the comprehensive characteristic value, the more significant the abnormal change of the corresponding position temperature and the dynamic abnormal change of the motor thermal state under the influence of different dominant factors, and the greater the influence of the temperature monitoring change on the motor thermal management control.

[0064] Based on the analysis results of the influence of motor thermal state change characteristics on different position temperatures under thermal management control, the thermal management control of brushless direct current motor during operation is dynamically adjusted, specifically:

[0065] The exponential smoothing algorithm is used to predict and analyze the temperature collected at each position, wherein the attenuation parameter of temperature prediction at each position is determined by the comprehensive characteristic value of each position temperature data: the characteristic value of the abnormal change of the temperature dynamic correlation of all positions is taken as the input, the Softmax function is used to obtain the normalization processing result of the characteristic value of the abnormal change of the temperature dynamic correlation of each position, and the normalization processing result is taken as the attenuation parameter of each temperature prediction analysis, that is, the abnormal correlation characteristics of the temperature change and the motor thermal state change of different positions under different dominant factors are considered to determine the attenuation parameter. If the characteristic value of the abnormal correlation is greater, it means that the abnormality of the current position recent data is significantly related to the abnormal change of the motor thermal state, and a larger attenuation parameter is set to improve the accuracy of monitoring and feedback of temperature abnormality of different positions in the process of motor thermal management.

[0066] Further, the temperature prediction result of each temperature data in the operation process of the brushless direct current motor is obtained by using an exponential smoothing algorithm based on the determined attenuation parameters of each temperature data, the difference between the temperature prediction result and the motor operation temperature threshold is calculated respectively, the maximum value of all the differences is taken as the feedback signal of the PID controller of the cooling system corresponding to the temperature control, and then the control signal of the current brushless direct current motor for temperature control by the cooling system is obtained through the PID controller, the control signal is transmitted to the cooling system to adjust the flow, and the temperature optimization control in the thermal management process is realized. Wherein, the exponential smoothing algorithm and the PID control are all known technologies, and the specific process will not be repeated.

[0067] The acquisition process of the correlation characteristic factor is shown in Figure 2

[0068] In summary, the embodiments of the present application adopt a multi-sensor network to adopt a hierarchical deployment and distributed collection mode to collect state monitoring data in the motor thermal management process, further analyze the temperature dynamic abnormal correlation characteristics of the motor at different positions under the influence of different dominant factors, based on the analysis result, more accurately analyze the temperature abnormality of the motor, and then determine the accurate temperature deviation of the motor monitoring, accurately feedback the control under the water cooling and cooling treatment according to the temperature deviation, improve the temperature control in the thermal management process of the brushless direct current motor, accurately implement the motor thermal management treatment, and further improve the stability of the brushless direct current motor operation.

[0069] It should be noted that the above-mentioned sequence of the embodiments of the present application is only for description, and does not represent the advantages and disadvantages of the embodiments. And the above describes the specific embodiments of the present application. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multi-task processing and parallel processing are also possible or may be advantageous.

[0070] Each embodiment in the present application is described in a progressive manner, and the same or similar parts of each embodiment can be referred to each other, and each embodiment mainly describes the difference from other embodiments.

[0071] The above-described embodiments are only used to illustrate the technical solutions of the present application, but not to limit them; the technical solutions described in the above embodiments are modified, or some technical features are replaced, without changing the essence of the corresponding technical solutions out of the scope of the technical solutions of the embodiments of the present application, which should be included in the protection scope of the present application.​

Claims

1. A method for optimizing thermal management control of brushless DC motors based on multi-sensor networks, characterized in that, The method includes the following steps: Real-time acquisition of stator phase current, motor speed, and temperature data at various locations of the brushless DC motor; Anomaly scores for each type of data at each time point are obtained through anomaly detection algorithms, and anomaly score sequences for each type of data are constructed. Stator phase current and motor speed are used as feature parameters. Based on the differences between each temperature data and the anomaly score sequences of all types of feature parameters, as well as the overall distribution of all anomaly scores, the correlation feature factors of motor thermal state anomalies for each temperature data are constructed. Combined with the anomaly scores of various temperature data at each time point, the first feature coefficients for each time point are constructed. Based on the first feature coefficient, time is divided. In each time period, according to the correlation between each temperature data and the time series of all other types of temperature data, and the similarity between the first feature coefficient at all times and all anomaly scores of each temperature data, the temperature dynamic correlation feature value of each temperature data in each time period is constructed. Based on the dynamic temperature correlation feature value and the correlation feature factor, a comprehensive feature value of various temperature data is constructed, the attenuation parameter for various temperature predictions is determined, and combined with the prediction algorithm, various temperatures are predicted for motor thermal management control.

2. The brushless DC motor thermal management control optimization method based on a multi-sensor network as described in claim 1, characterized in that, The process of obtaining the associated feature factors is as follows: Calculate the Euclidean distance between each feature parameter and the anomaly score sequence for each temperature data; calculate the fusion value of the anomaly scores for each feature parameter and each temperature data at all time points; The abnormal correlation between each feature parameter and each temperature data is positively correlated with the Euclidean distance and the fusion value, respectively. The sum of the abnormal correlations between each temperature data point and all types of feature parameters is used as the correlation feature value for each temperature data point.

3. The brushless DC motor thermal management control optimization method based on a multi-sensor network as described in claim 2, characterized in that, The fusion value of the anomaly score is the sum of the mean anomaly score for each feature parameter at all times and the mean anomaly score for each temperature data at all times.

4. The brushless DC motor thermal management control optimization method based on a multi-sensor network as described in claim 2, characterized in that, The abnormal correlation degree is the product of the Euclidean distance and the fusion value.

5. The brushless DC motor thermal management control optimization method based on a multi-sensor network as described in claim 1, characterized in that, The process of obtaining the first feature coefficient is as follows: The product of the correlation feature factor of each temperature data and the anomaly score of each temperature data at each time is recorded as the first product; the sum of the first products of all types of temperature data at each time is used as the first feature coefficient at each time.

6. The brushless DC motor thermal management control optimization method based on a multi-sensor network as described in claim 1, characterized in that, The method for dividing time is as follows: The first characteristic coefficients of all times are sorted in chronological order and curve fitting is performed. The slope of the corresponding data point in the fitted curve at each time point is calculated. The time points where the slope is not equal to 0 and the first characteristic coefficient of the corresponding time point is equal to 1 are used as the dividing points to obtain each time period.

7. The brushless DC motor thermal management control optimization method based on a multi-sensor network as described in claim 1, characterized in that, The process for obtaining the dynamic temperature correlation feature value is as follows: Within each time period, the mean correlation between each type of temperature data and the time series of all other types of temperature data is calculated using a similarity algorithm, and is denoted as the first response value; the similarity between the set of first feature coefficients at all times in the time period and the set of abnormal scores for each type of temperature data at all times is calculated, and is denoted as the second response value. The temperature dynamic correlation feature value of each temperature data in each time period is positively correlated with the second response value and negatively correlated with the first response value.

8. The brushless DC motor thermal management control optimization method based on a multi-sensor network as described in claim 7, characterized in that, The temperature dynamic correlation characteristic value is the result of calculating the ratio by using the sum of the first response value and the preset adjustment parameter as the denominator and the second response value as the numerator.

9. The brushless DC motor thermal management control optimization method based on a multi-sensor network as described in claim 1, characterized in that, The comprehensive feature value is the product of the mean of the temperature dynamic correlation feature values ​​for each temperature data across all time periods and the correlation feature factor for each temperature data.

10. The brushless DC motor thermal management control optimization method based on a multi-sensor network as described in claim 1, characterized in that, The attenuation parameter for various temperature predictions is the normalized value of the comprehensive characteristic value of various temperature data.

Images