Motor cooling system real-time intelligent control method based on edge computing

Abstract

The application discloses a motor cooling system real-time intelligent control method based on edge computing and relates to the technical field of motor cooling system control, which comprises the following steps: collecting the temperature and current of a water-cooled motor, collecting the flow and temperature of the cooling liquid, performing first screening processing and second screening processing to obtain current temperature flow change data, establishing a temperature change model of the water-cooled motor according to the current temperature flow change data, collecting corresponding current, temperature and flow during the operation of the water-cooled motor, performing temperature accurate analysis, and performing real-time control on the cooling system. The application is used to solve the problem that the existing motor cooling system control technology cannot estimate the change of the motor temperature according to the change of the load current, the cooling liquid temperature and the flow when controlling the flow of the cooling liquid of the water-cooled motor, and then judge the collected motor temperature in real time and control the flow of the cooling liquid in real time.

Description

Technical Field

[0001] This invention relates to the field of motor cooling system control technology, specifically to a real-time intelligent control method for motor cooling systems based on edge computing. Background Technology

[0002] Motor cooling system control technology is a comprehensive control technology with the core objective of maintaining the motor temperature within safe limits. It is a closed-loop system composed of a sensing and monitoring unit, a control decision unit, and an execution and adjustment unit. It adjusts the flow rate, velocity, temperature, and heat exchange path of the cooling medium in real time, dynamically, and on demand, and integrates anomaly diagnosis and redundant protection. It also takes into account cooling efficiency, energy consumption, and noise control.

[0003] Existing motor cooling system control technologies typically control the coolant flow rate of water-cooled motors by collecting the temperature of the stator windings using sensors, and then controlling the coolant flow rate in real time based on this temperature. However, the stator windings are the core heat-generating area of ​​a water-cooled motor, but their extreme environment—characterized by strong electromagnetic fields, high vibration, humid and hot conditions, and large temperature gradients—makes sensors highly susceptible to noise. This noise is not random error but rather multi-source, clearly defined interference signals, directly leading to misjudgments in coolant flow control and frequent fluctuations in pump speed. For example, patent application CN117519330A discloses a control method, a motor cooling system, and a motor for a motor cooling system. This solution controls the coolant flow rate by collecting the temperature of the stator windings. The current control system, which does not perform noise assessment on the collected temperature data, is prone to erroneous control. To avoid this, existing methods often collect temperature data for a period of time after detecting suspected noise, combining the data for judgment. However, this second data collection and judgment requires setting a fixed delay window. During the delay, the system can only maintain the current flow rate, which can lead to untimely control of the cooling system. In certain special scenarios, this can easily cause the motor to exceed the safe temperature. Therefore, existing motor cooling system control technology cannot predict changes in motor temperature based on changes in load current, coolant temperature, and flow rate when controlling the coolant flow rate of water-cooled motors, thus failing to make real-time judgments on the collected motor temperature and control the coolant flow rate in real time. Summary of the Invention

[0004] This invention aims to at least partially solve one of the technical problems in the prior art. It obtains motor temperature and current data and water cooling temperature and flow rate data by collecting the temperature and current of a water-cooled motor and the flow rate and temperature of the coolant, respectively. After performing a first screening process and a second screening process, it obtains current, temperature, and flow rate change data. Based on these data, it establishes a temperature change model for the water-cooled motor. During the operation of the water-cooled motor, it collects the corresponding current, temperature, and flow rate, performs accurate temperature analysis based on the motor temperature change model, and controls the cooling system in real time. This addresses the problem that existing motor cooling system control technologies cannot predict motor temperature changes based on variations in load current, coolant temperature, and flow rate when controlling the coolant flow rate of a water-cooled motor, thus failing to accurately judge the collected motor temperature and control the coolant flow rate in real time.

[0005] To achieve the above objectives, this application provides a real-time intelligent control method for a motor cooling system based on edge computing, comprising the following steps:

[0006] The temperature and current of the water-cooled motor were collected, as well as the flow rate and temperature of the coolant, to obtain motor temperature and current data and water cooling temperature and flow rate data, respectively.

[0007] The motor temperature data and water cooling temperature and flow rate data are subjected to a first screening process and a second screening process to obtain the current temperature and flow rate change data.

[0008] Based on a multilayer sensor, and by establishing a temperature change model for the water-cooled motor according to current, temperature and flow rate change data, the motor temperature change model is obtained.

[0009] During the operation of the water-cooled motor, the corresponding current, temperature and flow rate are collected, and the temperature is accurately analyzed based on the motor temperature change model, and the cooling system is controlled in real time.

[0010] Furthermore, the temperature and current of the water-cooled motor are collected, as well as the flow rate and temperature of the coolant, to obtain motor temperature and current data and water cooling temperature and flow rate data, including the following sub-steps:

[0011] Let any one of the water-cooled motors be referred to as the first motor, let the temperature of the stator winding of the first motor be referred to as the motor temperature of the first motor, and let the coolant of the first motor be referred to as the water coolant.

[0012] The temperature of the coolant at the inlet is recorded as the inlet temperature, and the temperature of the coolant at the outlet is recorded as the outlet temperature. The difference between the inlet temperature and the outlet temperature is recorded as the coolant temperature difference. The flow rate at the coolant inlet is recorded as the coolant flow rate.

[0013] Furthermore, collecting the temperature and current of the water-cooled motor, and the flow rate and temperature of the coolant, to obtain the motor temperature and current data and the water cooling temperature and flow rate data respectively, also includes the following sub-steps:

[0014] During the normal operation of the first motor, the motor temperature and load current of the first motor are synchronously collected at a first time interval, and the collection time is recorded. The data are arranged in chronological order and recorded as motor temperature sequence and load current sequence, respectively, and labeled as motor temperature and current data. The first time interval is t1.

[0015] Simultaneously, the coolant temperature difference and coolant flow rate are collected at the first time interval, and the collection time is recorded. The data are then arranged in chronological order and recorded as the coolant temperature difference sequence and coolant flow rate sequence, respectively, and labeled as coolant temperature and flow rate data.

[0016] Further, the motor temperature data and water cooling temperature and flow rate data undergo a first screening process and a second screening process to obtain the current, temperature, and flow rate change data, including the following sub-steps:

[0017] Let any motor temperature in the motor temperature sequence be denoted as the first temperature DT. i , where i represents the position number; calculate the first-order difference sequence corresponding to the motor temperature sequence, and calculate the corresponding second-order difference sequence based on the corresponding first-order difference sequence, and then calculate the corresponding third-order difference sequence based on the corresponding second-order difference sequence.

[0018] DT i The corresponding third-order difference is denoted as D3T i Calculate D3T i-e1 To D3T i-1 The average value, denoted as TP i And calculate |TP i |×2, denoted as D3T i The corresponding constraint threshold; if |D3T i If the value is greater than the corresponding constraint threshold, then DT will be... i Mark it as the first suspicious value, where e1 is the number to be set.

[0019] Furthermore, the process of performing a first screening and a second screening on the motor temperature data and water cooling temperature and flow rate data to obtain the current, temperature, and flow rate change data includes the following sub-steps:

[0020] With DT i Take the distance DT in the motor temperature sequence as the center. i The recent e2 motor temperatures, compared to DT i Composition of DT i The neighborhood window is used to sort the motor temperatures in the neighborhood window in order of magnitude, and the corresponding median is obtained, denoted as WT.i Where e2 is the number to be set;

[0021] Calculate 1 / (1+|DT) i -WT i |), denoted as DT i Density aggregation degree MP i Repeatedly calculate the density clustering degree of all motor temperatures in the neighborhood window, and calculate the mean RM and standard deviation BM of all density clustering degrees; if MP i If it is not located in [RM-k1×BM, RM+k1×BM], then DT will be... i Marked as the second suspicious value, where k1 is the set proportional coefficient;

[0022] If DT i If it is neither the first nor the second suspicious value, then DT will be... i Mark as normal value, otherwise DT i The abnormal values ​​of motor temperature are marked, and the corresponding acquisition time is recorded as the abnormal time.

[0023] Repeatedly filter out all outliers and their corresponding outlier times in the motor temperature sequence, and repeat the same process for filtering out all outliers and their corresponding outlier times in the load current sequence and coolant flow sequence.

[0024] Furthermore, the process of performing a first screening and a second screening on the motor temperature data and water cooling temperature and flow rate data to obtain the current, temperature, and flow rate change data includes the following sub-steps:

[0025] Let any one of the cold liquid temperature differences in the cold liquid temperature difference sequence be denoted as the first temperature difference RC. j Where j represents the position number, if RC j If not greater than 0, then RC j Mark it as an outlier; otherwise mark it as a suspected outlier.

[0026] If RC j If it is a suspected value, then RC j Take the distance RC in the cold liquid temperature difference sequence j The most recent e3 motor temperatures, denoted as RC j The nearest value, with RC j RC j The adjacent windows, where e3 is the set number;

[0027] Get the median WR of all neighboring values, and calculate the absolute difference between each neighboring value and WR, denoted as the neighboring median difference. Calculate the mean PG and standard deviation BG of all neighboring medians. The neighboring values ​​corresponding to the neighboring medians that are not located in [PG-k2×BG, PG+k2×BG] are recorded as abnormal neighboring values ​​and removed from the neighboring window, where k2 is the set scaling factor.

[0028] Divide the remaining adjacent windows into two equal parts in sequence, and denote them as the front window and the back window respectively. Calculate the standard deviations of the front window and the back window respectively, and denote them as VL and VR in sequence.

[0029] Calculate min(VL, VR) / max(VL, VR), denoted as the fluctuation conservation coefficient SV. If SV is less than k3, then RC... j The abnormal values ​​of the cold liquid temperature difference are marked and the corresponding acquisition time is recorded as the abnormal time. Otherwise, they are marked as normal values. The process of filtering out all abnormal values ​​and their corresponding abnormal times in the cold liquid temperature difference sequence is repeated, where k3 is the set threshold.

[0030] Furthermore, the process of performing a first screening and a second screening on the motor temperature data and water cooling temperature and flow rate data to obtain the current, temperature, and flow rate change data includes the following sub-steps:

[0031] For the motor temperature sequence, load current sequence, coolant flow rate sequence, and coolant temperature difference sequence, remove the motor temperature, load current, coolant flow rate, and coolant temperature difference collected at any abnormal moment from the corresponding sequence; after completion, the first temperature sequence, first current sequence, first flow rate sequence, and first temperature difference sequence are obtained respectively.

[0032] The first temperature sequence is divided into multiple subsequences using the abnormal time as the dividing position. Any subsequence is denoted as the first subsequence. The first difference sequence corresponding to the first subsequence is calculated and denoted as the first difference sequence.

[0033] Let any first difference in the first difference numerator sequence be denoted as FC. m Where m represents the position number, taken as the distance FC in the first temperature sequence. m The most recent e4 data points, where e4 is the number of settings; denoted as FC. m Reference data, obtain all reference data and FC m The median FM.

[0034] Furthermore, the process of performing a first screening and a second screening on the motor temperature data and water cooling temperature and flow rate data to obtain the current, temperature, and flow rate change data includes the following sub-steps:

[0035] Calculate the absolute difference between all reference data and FM, and obtain the median CM of all absolute differences; if |FC m If -FM| is greater than k4×CM, then FC will be applied. m This is recorded as an abnormal difference, and the two corresponding acquisition times are also marked as abnormal times, where k4 is the set scaling factor;

[0036] Repeatedly screen all abnormal differences and corresponding abnormal times in the first difference sequence to obtain the corresponding screening subsequence. Repeatedly obtain the screening subsequences corresponding to all subsequences to obtain the screening difference sequence corresponding to the first temperature sequence.

[0037] Repeatedly obtain the filtered differential sequences corresponding to the current sequence, the first flow rate sequence, and the first temperature difference sequence, and remove all first-order differences involved in any abnormal moment from the corresponding filtered differential sequences. After completion, the differential temperature sequence, differential current sequence, differential flow rate sequence, and differential temperature difference sequence are obtained respectively, and are recorded as current, temperature and flow rate change data.

[0038] Furthermore, based on a multilayer perceptron and using data on changes in current, temperature, and flow rate, a temperature change model for the water-cooled motor is established. This process further includes the following sub-steps:

[0039] The first-order differences in the differential temperature sequence, differential current sequence, differential flow rate sequence, and differential temperature difference sequence are respectively denoted as differential temperature, differential current, differential flow rate, and differential temperature difference.

[0040] An initial model is constructed based on a multilayer perceptron. The initial model includes an input layer, a hidden layer, and an output layer. The input of the initial model is set to the differential current, differential flow rate, and differential temperature difference at the same acquisition time, and the output is the differential temperature.

[0041] The data on changes in current, temperature, and flow rate are divided into a training set and a test set. The initial model is trained using the training set to obtain the motor temperature change model. The motor temperature change model is then tested using the test set. The mean absolute error of the motor temperature change model is calculated and denoted as MAE.

[0042] Furthermore, during the operation of the water-cooled motor, the corresponding current, temperature, and flow rate are collected, and accurate temperature analysis is performed based on the motor temperature change model. Real-time control of the cooling system includes the following sub-steps:

[0043] During normal operation of the first motor, the motor temperature, load current, coolant temperature difference and coolant flow rate of the first motor are synchronously collected at a first time interval; and the differential temperature ET, differential current, differential flow rate and differential temperature difference corresponding to the current collection time are calculated.

[0044] The differential current, differential flow rate, and differential temperature difference are input into the motor temperature change model to obtain the differential temperature output by the model, which is denoted as the theoretical temperature difference LT. If ET is not located in [k5×(LT-MAE), k5×(LT+MAE)], the motor temperature collected at the current acquisition time is marked as the noise temperature, and the flow rate of the coolant is not controlled. Here, k5 is the set proportional coefficient.

[0045] If ET is located in [k5×(LT-MAE), k5×(LT+MAE)], then the motor temperature collected at the current acquisition time is marked as the accurate temperature, and the flow rate of the coolant is controlled by PID based on the current acquired motor temperature.

[0046] The beneficial effects of this invention are as follows: This invention collects the temperature and current of a water-cooled motor, and the flow rate and temperature of the coolant, to obtain motor temperature and current data and water cooling temperature and flow rate data, respectively; it performs a first screening process and a second screening process on the motor temperature data and water cooling temperature and flow rate data to obtain current, temperature, and flow rate change data; based on a multilayer sensor, and according to the current, temperature, and flow rate change data, it establishes a temperature change model for the water-cooled motor, thus obtaining a motor temperature change model; during the operation of the water-cooled motor, it collects the corresponding current, temperature, and flow rate, and performs accurate temperature analysis based on the motor temperature change model, and controls the cooling system in real time; when controlling the flow rate of the coolant in the water-cooled motor, it can predict the change in motor temperature based on the changes in load current, coolant temperature, and flow rate, and then judge the collected motor temperature in real time, improving the accuracy and timeliness of control;

[0047] This invention uses the third-order difference to construct a constraint threshold to determine the first suspicious value, thus identifying genuine spike noise instead of misinterpreting it as a slow trend and avoiding mistaking short-term noise for normal data. It also uses the median in a neighborhood window, calculates density clustering, and uses the mean and variance to determine the second suspicious value, effectively identifying long-term shifts or isolated anomalies while preserving local real changes. For the cold liquid temperature difference, it uses a neighboring window and median difference to remove anomalies, and calculates the fluctuation conservation coefficient SV by dividing the window into two parts to determine if it is normal. Furthermore, based on physical laws, if the cold and hot flows do not satisfy the basic fluctuation symmetry or continuity in the time series, it may be due to sensor errors or sudden changes in operating conditions, accurately eliminating noise points in the cold liquid temperature difference. Attached Figure Description

[0048] Figure 1 This is a flowchart illustrating the steps of the method of the present invention;

[0049] Figure 2 This is a flowchart for screening outlier values ​​in motor temperature according to the present invention.

[0050] Figure 3 This is a flowchart of the outlier screening process for the cold liquid temperature difference according to the present invention.

[0051] Figure 4 This is a schematic diagram of the electronic device of the present invention. Detailed Implementation

[0052] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0053] Example 1, please refer to Figure 1 As shown, this application provides a real-time intelligent control method for a motor cooling system based on edge computing, including the following steps:

[0054] Step S1 involves collecting the temperature and current of the water-cooled motor, as well as the flow rate and temperature of the coolant, to obtain motor temperature and current data and water cooling temperature and flow rate data, respectively. Step S1 includes the following sub-steps:

[0055] Step S101: Any water-cooled motor is designated as the first motor, the temperature of the stator winding of the first motor is designated as the motor temperature of the first motor, and the coolant of the first motor is designated as the water coolant; the copper loss of the stator winding is the main heat source of the water-cooled motor, and its temperature can represent the motor temperature.

[0056] Step S102: Record the temperature of the coolant at the inlet as the inlet temperature, and the temperature of the coolant at the outlet as the outlet temperature. Record the difference between the inlet temperature and the outlet temperature as the coolant temperature difference; that is, coolant temperature difference = inlet temperature - outlet temperature. Record the flow rate at the coolant inlet as the coolant flow rate.

[0057] Step S103: During the normal operation of the first motor, the motor temperature and load current of the first motor are synchronously collected at a first time interval, and the collection time is recorded. The data are arranged in chronological order and recorded as motor temperature sequence and load current sequence, respectively, and marked as motor temperature and current data. The first time interval is t1. In this embodiment, the first time interval is 0.1s. The sampling frequency can be selected according to the actual application scenario.

[0058] Step S104, and simultaneously collect the coolant temperature difference and coolant flow rate at the first time interval, record the collection time, and arrange them in chronological order, respectively as coolant temperature difference sequence and coolant flow rate sequence, and mark them as coolant temperature and flow rate data;

[0059] In practical implementation, the load current is the core heat source of the motor. An increase in current directly increases the heat generation power of copper loss, iron loss, etc., and drives the motor temperature to rise; a decrease in current reduces the heat generation power and causes the temperature to drop. Changes in coolant flow rate and coolant temperature difference together characterize changes in heat exchange efficiency. An increase in flow rate or a positive change in temperature difference will increase the heat exchange power of the coolant, accelerate the removal of heat from the motor, and inhibit the rise in motor temperature; conversely, a decrease in flow rate or temperature difference will reduce the heat exchange power and accelerate the rise in motor temperature.

[0060] Step S2 involves performing a first filtering process and a second filtering process on the motor temperature data and water cooling temperature and flow rate data to obtain current, temperature, and flow rate change data. Step S2 includes the following sub-steps:

[0061] For step S201, please refer to... Figure 2 As shown, any motor temperature in the motor temperature sequence is denoted as the first temperature DT. i , where i represents the position number; calculate the first-order difference sequence corresponding to the motor temperature sequence, and calculate the corresponding second-order difference sequence based on the corresponding first-order difference sequence, and then calculate the corresponding third-order difference sequence based on the corresponding second-order difference sequence; for example, DT i First-order difference D1T i =DT i -DT i-1 Second-order difference D2T i =D1T i -D1T i-1 Third-order difference D3T i =D2T i -D2T i-1 ;

[0062] Step S202, DT i The corresponding third-order difference is denoted as D3T i Calculate D3T i-e1 To D3T i-1 The average value if D3T i If the number of third-order differences is less than e1, then start from D3T. i Then, complete the missing parts, denoted as TP. i And calculate |TP i |×2, denoted as D3T i The corresponding constraint threshold; if |D3T i If the value is greater than the corresponding constraint threshold, then DT will be... i The value is marked as the first suspicious value, where e1 is the number of values ​​to be set; in this embodiment, e1=4, but it can be set flexibly, generally from 4 to 7.

[0063] Third-order difference emphasizes higher-order abrupt changes and pulses in the sequence. For occasional sharp pulses from the sensor, first-order and second-order differences may not be significant, but third-order difference amplifies this short-term discontinuity, making it easier to capture transient noise. By using the local mean of several past third-order differences, the threshold can be dynamically set to adapt to the local fluctuation amplitude of the sequence and avoid missed detection under different operating conditions.

[0064] Step S203, with DT i Take the distance DT in the motor temperature sequence as the center. i The recent e2 motor temperatures, compared to DT i Composition of DT i The neighborhood window is used to sort the motor temperatures in the neighborhood window in order of magnitude, and the corresponding median is obtained, denoted as WT. i , where e2 is the number to be set; in this embodiment, e2=8, which can be set flexibly, generally from 6 to 10;

[0065] Step S204, calculate 1 / (1+|DT) i -WT i |), denoted as DT i Density aggregation degree MP i Repeatedly calculate the density clustering degree of all motor temperatures in the neighborhood window, and calculate the mean RM and standard deviation BM of all density clustering degrees; if MP i If it is not located in [RM-k1×BM, RM+k1×BM], then DT will be... i The value is marked as the second suspicious value, where k1 is the set proportional coefficient; in this embodiment, k1=2, which can be set flexibly, generally [2, 3]; if a sampled value deviates far from the median of its neighborhood, |DTi-WTi| will be large, resulting in a small density cluster, and thus it will be identified as an anomaly; using the median instead of the mean makes this discrimination more robust to existing anomalies in the neighborhood;

[0066] For example, if the motor temperature sequence is {90.9, 91.1, 91.3, 91.5, 105.0, 91.8, 92.0, 92.2, 92.4}, in °C, then the median is 91.8, and the corresponding density clustering degree is {0.526, 0.588, 0.667, 0.769, 0.070, 1.000, 0.833, 0.714, 0.625}. Then RM=0.644, BM=0.258. Then [RM-k1×BM, RM+k1×BM] is [0.128, 1, 16]. The density clustering degree corresponding to 105.0 is not in this range, so 105.0 is the second suspicious value.

[0067] Step S205, if DT i If it is neither the first nor the second suspicious value, then DT will be...i Mark as normal value, otherwise DT i The abnormal values ​​of motor temperature are marked, and the corresponding acquisition time is recorded as the abnormal time.

[0068] Step S206: Repeatedly filter out all abnormal values ​​and corresponding abnormal times in the motor temperature sequence, and repeat the same process for filtering out all abnormal values ​​and corresponding abnormal times in the load current sequence and coolant flow sequence; record the abnormal times for easy removal later.

[0069] For step S207, please refer to... Figure 3 As shown, any one of the cold liquid temperature differences in the cold liquid temperature difference sequence is denoted as the first temperature difference RC. j Where j represents the position number, if RC j If not greater than 0, then RC j If a value is marked as an outlier, logically the coolant inlet temperature should not be lower than the outlet temperature; otherwise, it should be marked as a suspected value.

[0070] Step S208, if RC j If it is a suspected value, then RC j Take the distance RC in the cold liquid temperature difference sequence j The most recent e3 motor temperatures, denoted as RC j The nearest value, with RC j RC j The adjacent windows, where e3 is the number to be set; in this embodiment, e3=5, which can be set flexibly;

[0071] Step S209: Obtain the median WR of all neighboring values, and calculate the absolute difference between each neighboring value and WR, denoted as the neighboring median difference. Calculate the average PG and standard deviation BG of all neighboring median differences. Record the neighboring values ​​corresponding to the neighboring median differences that are not located in [PG-k2×BG, PG+k2×BG] as abnormal neighboring values ​​and remove them from the neighboring window. Here, k2 is a set proportional coefficient. In this embodiment, k2=2, which can be set flexibly. Generally, it is [2, 3]. Removing neighboring values ​​that deviate greatly from the median first can avoid the influence of other possible outliers.

[0072] Step S210: Divide the remaining adjacent windows into two parts in sequence, and denote them as the front window and the back window respectively. Calculate the standard deviations of the front window and the back window respectively, and denote them as VL and VR in sequence.

[0073] Step S211: Calculate min(VL, VR) / max(VL, VR), denoted as the fluctuation conservation coefficient SV. If SV is less than k3, then RC... jThe abnormal values ​​of the cold liquid temperature difference are marked as abnormal values ​​and the corresponding collection time is recorded as abnormal time. Otherwise, they are marked as normal values. The process of filtering out all abnormal values ​​and their corresponding abnormal times in the cold liquid temperature difference sequence is repeated. Here, k3 is the set threshold. In this embodiment, k3=0.6. It can be flexibly set according to the actual application scenario. Generally, it is [0.5, 0.7].

[0074] For example, RC j =0.05, neighboring values ​​are {0.12, 0.11, 0.09, 0.10, 0.08}, median WR=0.10, neighboring median differences are {0.02, 0.01, 0.01, 0.0, 0.02}, mean PG=0.012, standard deviation BG=0.007, all within [PG-k2×BG, PG+k2×BG]; therefore, the neighboring window is {0.12, 0.11, 0.09, 0.05, 0.10, 0.08}, the VL of the first window is 0.012, the VR of the second window is 0.021, then SV=0.012 / 0.021=0.57, which is less than 0.6, therefore RC j This is an outlier;

[0075] The temperature difference of the coolant is a direct reflection of the heat exchange capacity and should meet the physical continuity of fluctuation. In the absence of systemic failures, the fluctuation amplitude of the coolant temperature difference should be similar before and after in a short neighborhood. The SV test can detect abrupt changes, directional changes, or anomalies masked by single-point interference, which is more accurate than simply looking at the magnitude.

[0076] Step S212: For the motor temperature sequence, load current sequence, coolant flow rate sequence, and coolant temperature difference sequence, remove the motor temperature, load current, coolant flow rate, and coolant temperature difference collected at any abnormal moment from the corresponding sequence; after completion, the first temperature sequence, first current sequence, first flow rate sequence, and first temperature difference sequence are obtained respectively.

[0077] Step S213: Divide the first temperature sequence into multiple subsequences using the abnormal time as the dividing position, and denote any one of the subsequences as the first subsequence. Calculate the first-order difference sequence corresponding to the first subsequence and denote it as the first difference sequence. Dividing into subsequences is to avoid the impact of data removed at abnormal times on the difference statistics. Calculating the difference within each continuous and relatively clean segment is more accurate.

[0078] Step S214: Denote any first-order difference in the first difference numerator sequence as FC. m Where m represents the position number, taken as the distance FC in the first temperature sequence. m The most recent e4 data points, where e4 is the number of settings; denoted as FC. m Reference data, obtain all reference data and FC mIf the median FM is given, then in this embodiment, e4=4, which can be set flexibly; generally it is [4, 6].

[0079] Step S215: Calculate the absolute difference between all reference data and FM, and obtain the median CM of all absolute differences; if |FC m If -FM| is greater than k4×CM, then FC will be applied. m This is recorded as an abnormal difference, and the two corresponding acquisition times are also marked as abnormal times. Here, k4 is a set scaling factor; in this embodiment, k4=2, which can be flexibly set. CM accurately represents the true fluctuation range of normal change values ​​within the window; if |FC m -FM| is greater than k4×CM, i.e., FC m If the fluctuation range is abnormal, it indicates that FC m This is an outlier;

[0080] For example, FC m =5.0, the corresponding reference data is {0.2, 0.3, 5.0, 0.2, 0.3}, then the median FM = 0.3, the corresponding absolute difference is {0.1, 0, 4.7, 0.1, 0}, then CM = 0.1, k4 × CM = 0.2, then |FC m If -FM|=4.7>0.2, then FC m This is an abnormal difference.

[0081] Step S216: Repeatedly screen all abnormal differences and corresponding abnormal times in the first difference sequence to obtain the corresponding screening subsequences. Repeatedly obtain the screening subsequences corresponding to all subsequences to obtain the screening difference sequence corresponding to the first temperature sequence.

[0082] Step S217: Repeatedly acquire the filtered difference sequences corresponding to the current sequence, the first flow rate sequence, and the first temperature difference sequence, and remove all first-order differences involved in any abnormal moment from the corresponding filtered difference sequences. After completion, the differential temperature sequence, differential current sequence, differential flow rate sequence, and differential temperature difference sequence are obtained respectively, and recorded as current, temperature, and flow rate change data. When a certain moment is judged to be abnormal, in order to ensure data consistency, the data collected and calculated at that moment need to be removed at the same time to avoid missing paired features or the model learning incorrect variable relationships during subsequent model training.

[0083] In the specific implementation process, the motor temperature sequence, load current sequence, and coolant flow sequence are directly acquired by sensors, have clear absolute physical limits, and are continuous and have inertia; while the coolant temperature difference sequence is calculated from two directly acquired quantities, without independent sensors, and anomalies come not only from the original temperature sensors, but also from errors in the calculation logic, etc., and the physical constraints are relative limits. Therefore, different outlier screening methods should be used. If a uniform method is used, a large number of misjudgments or omissions will occur due to the differences in the sequences.

[0084] Step S3: Based on a multilayer sensor and using current, temperature, and flow rate change data, a temperature change model for the water-cooled motor is established to obtain the motor temperature change model. Step S3 includes the following sub-steps:

[0085] Step S301: The first-order differences in the differential temperature sequence, differential current sequence, differential flow rate sequence, and differential temperature difference sequence are respectively denoted as differential temperature, differential current, differential flow rate, and differential temperature difference.

[0086] Step S302: Construct an initial model based on a multilayer perceptron. The initial model includes an input layer, a hidden layer, and an output layer. Set the input of the initial model to the differential current, differential flow rate, and differential temperature difference at the same acquisition time, and set the output to the differential temperature.

[0087] Step S303: Divide the current, temperature and flow rate change data into training set and test set. Use the training set to train the initial model. After training, the motor temperature change model is obtained. Use the test set to test the motor temperature change model. Calculate the mean absolute error of the motor temperature change model, denoted as MAE.

[0088] In practical implementation, the motor temperature change model has three input dimensions and one output dimension. It does not require a deep network, has extremely low model complexity, is fast in training and inference, and is not easily fitted. It can be deployed on water-cooled motors for edge control.

[0089] Step S4 involves collecting the corresponding current, temperature, and flow rate during the operation of the water-cooled motor, performing accurate temperature analysis based on the motor temperature change model, and controlling the cooling system in real time. Step S4 includes the following sub-steps:

[0090] Step S401: During the normal operation of the first motor, the motor temperature, load current, and coolant temperature difference and coolant flow rate of the first motor are synchronously collected at a first time interval; and the differential temperature ET, differential current, differential flow rate and differential temperature difference corresponding to the current collection time are calculated.

[0091] In step S402, the differential current, differential flow rate, and differential temperature difference are input into the motor temperature change model to obtain the differential temperature output by the model, denoted as the theoretical temperature difference LT. If ET is not located in [k5×(LT-MAE), k5×(LT+MAE)], the motor temperature collected at the current acquisition time is marked as the noise temperature, and the flow rate of the coolant is not controlled. Here, k5 is the set proportional coefficient. In this embodiment, k5=1.2, which can be flexibly set to [0.8, 1.3], allowing for appropriate adjustment. Using MAE to construct the interval incorporates the uncertainty of the model into the judgment, so as not to discard valuable measurements due to the error of the model itself or a small amount of residual noise.

[0092] Step S403: If ET is located in [k5×(LT-MAE), k5×(LT+MAE)], then the motor temperature collected at the current acquisition time is marked as the accurate temperature, and the flow rate of the coolant is controlled by PID according to the current acquired motor temperature; the control is only executed when the measurement is confirmed to be reliable, so as to avoid mechanical loss, energy consumption increase and system instability caused by repeated flow rate adjustment due to transient noise.

[0093] In practice, other control algorithms can be selected for the flow rate of the coolant depending on the actual application scenario.

[0094] Example 2, please refer to Figure 4 As shown, Figure 4 A schematic diagram of an electronic device is provided, which may include a processor, a communication interface, a memory, and a communication bus. The processor, communication interface, and memory communicate with each other via the communication bus. The memory stores computer-readable instructions, and the processor can call these instructions. When the processor executes a computer-readable instruction, it performs steps such as those in the real-time intelligent control method for a motor cooling system based on edge computing to achieve the following functions: collecting the temperature and current of the water-cooled motor, and collecting the flow rate and temperature of the coolant, obtaining motor temperature and current data and water cooling temperature and flow rate data, respectively; performing a first filtering process and a second filtering process on the motor temperature data and water cooling temperature and flow rate data to obtain current, temperature, and flow rate change data; establishing a temperature change model of the water-cooled motor based on a multilayer sensor and the current, temperature, and flow rate change data, obtaining a motor temperature change model; and collecting the corresponding current, temperature, and flow rate during the operation of the water-cooled motor, performing accurate temperature analysis based on the motor temperature change model, and controlling the cooling system in real time.

[0095] Furthermore, when the logical instructions in the aforementioned memory can be implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0096] Example 3: This application also provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it performs steps such as those in the real-time intelligent control method for a motor cooling system based on edge computing to achieve the following functions: collecting the temperature and current of the water-cooled motor, and collecting the flow rate and temperature of the coolant, to obtain motor temperature and current data and water cooling temperature and flow rate data, respectively; performing a first filtering process and a second filtering process on the motor temperature data and water cooling temperature and flow rate data to obtain current, temperature, and flow rate change data; establishing a temperature change model of the water-cooled motor based on a multilayer sensor and the current, temperature, and flow rate change data to obtain a motor temperature change model; during the operation of the water-cooled motor, collecting the corresponding current, temperature, and flow rate, performing accurate temperature analysis based on the motor temperature change model, and controlling the cooling system in real time.

[0097] Based on the above description of the embodiments, the embodiments of the present invention can be provided as methods, systems, or computer program products. Based on this understanding, the above technical solutions, in essence or in terms of their contribution to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or certain parts of the embodiments.

[0098] In the embodiments provided in this application, it should be understood that the disclosed system or method can be implemented in other ways. The embodiments described above are merely illustrative. For example, the division of modules or units is only a logical functional division, and there may be other division methods in actual implementation. Furthermore, multiple modules or units may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the coupling or direct coupling or communication connection shown or discussed may be through some communication interfaces. The indirect coupling or communication connection between systems, modules, and units may be electrical, mechanical, or other forms.

[0099] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A real-time intelligent control method for a motor cooling system based on edge computing, characterized in that, Includes the following steps: The temperature and current of the water-cooled motor were collected, as well as the flow rate and temperature of the coolant, to obtain motor temperature and current data and water cooling temperature and flow rate data, respectively. The motor temperature and current data, as well as the water cooling temperature and flow rate data, are subjected to a first screening process and a second screening process to obtain the current, temperature and flow rate change data. Based on a multilayer sensor, and by establishing a temperature change model for the water-cooled motor according to current, temperature and flow rate change data, the motor temperature change model is obtained. During the operation of the water-cooled motor, the corresponding current, temperature and flow rate are collected, and the temperature is accurately analyzed based on the motor temperature change model, and the cooling system is controlled in real time. The process of performing a first and second filtering process on motor temperature and current data, and water cooling temperature and flow rate data to obtain current, temperature, and flow rate change data includes the following sub-steps: Let any motor temperature in the motor temperature sequence be denoted as the first temperature DT. i , where i represents the position number; calculate the first-order difference sequence corresponding to the motor temperature sequence, and calculate the corresponding second-order difference sequence based on the corresponding first-order difference sequence, and then calculate the corresponding third-order difference sequence based on the corresponding second-order difference sequence. DT i The corresponding third-order difference is denoted as D3T i Calculate D3T i-e1 To D3T i-1 The average value, denoted as TP i And calculate |TP i |×2, denoted as D3T i The corresponding constraint threshold; if |D3T i If the value is greater than the corresponding constraint threshold, then DT will be... i Mark it as the first suspicious value, where e1 is the number of values ​​set; With DT i Take the distance DT in the motor temperature sequence as the center. i The recent e2 motor temperatures, compared to DT i Composition of DT i The neighborhood window is used to sort the motor temperatures in the neighborhood window in order of magnitude, and the corresponding median is obtained, denoted as WT. i Where e2 is the number to be set; Calculate 1 / (1+|DT) i -WT i |), denoted as DT i Density aggregation degree MP i Repeatedly calculate the density clustering degree of all motor temperatures in the neighborhood window, and calculate the mean RM and standard deviation BM of all density clustering degrees; If MP i If it is not located in [RM-k1×BM, RM+k1×BM], then DT will be... i Marked as the second suspicious value, where k1 is the set proportional coefficient; If DT i If it is neither the first nor the second suspicious value, then DT will be... i Mark as normal value, otherwise DT i The abnormal values ​​of motor temperature are marked, and the corresponding acquisition time is recorded as the abnormal time. Repeatedly filter out all outliers and their corresponding outlier times in the motor temperature sequence, and repeat the same process for filtering out all outliers and their corresponding outlier times in the load current sequence and coolant flow sequence.

2. The real-time intelligent control method for a motor cooling system based on edge computing according to claim 1, characterized in that, The process of collecting the temperature and current of the water-cooled motor, and the flow rate and temperature of the coolant, to obtain motor temperature and current data and water cooling temperature and flow rate data includes the following sub-steps: Let any one of the water-cooled motors be referred to as the first motor, let the temperature of the stator winding of the first motor be referred to as the motor temperature of the first motor, and let the coolant of the first motor be referred to as the water coolant. The temperature of the coolant at the inlet is recorded as the inlet temperature, and the temperature of the coolant at the outlet is recorded as the outlet temperature. The difference between the inlet temperature and the outlet temperature is recorded as the coolant temperature difference. The flow rate at the coolant inlet is recorded as the coolant flow rate.

3. The real-time intelligent control method for a motor cooling system based on edge computing according to claim 2, characterized in that, Collecting the temperature and current of the water-cooled motor, and collecting the flow rate and temperature of the coolant, to obtain motor temperature and current data and water cooling temperature and flow rate data, also includes the following sub-steps: During the normal operation of the first motor, the motor temperature and load current of the first motor are synchronously collected at a first time interval, and the collection time is recorded. The data are arranged in chronological order and recorded as motor temperature sequence and load current sequence, respectively, and labeled as motor temperature and current data. The first time interval is t1. Simultaneously, the coolant temperature difference and coolant flow rate are collected at the first time interval, and the collection time is recorded. The data are then arranged in chronological order and recorded as the coolant temperature difference sequence and coolant flow rate sequence, respectively, and labeled as coolant temperature and flow rate data.

4. The real-time intelligent control method for a motor cooling system based on edge computing according to claim 3, characterized in that, The process of performing first and second filtering on motor temperature and current data, and water cooling temperature and flow rate data to obtain current, temperature, and flow rate change data also includes the following sub-steps: Let any one of the cold liquid temperature differences in the cold liquid temperature difference sequence be denoted as the first temperature difference RC. j Where j represents the position number, if RC j If not greater than 0, then RC j Mark it as an outlier; otherwise mark it as a suspected outlier. If RC j If it is a suspected value, then RC j Take the distance RC in the cold liquid temperature difference sequence j The most recent e3 motor temperatures, denoted as RC j The nearest value, with RC j RC j The adjacent windows, where e3 is the set number; Get the median WR of all neighboring values, and calculate the absolute difference between each neighboring value and WR, denoted as the neighboring median difference. Calculate the mean PG and standard deviation BG of all neighboring medians. The neighboring values ​​corresponding to the neighboring medians that are not located in [PG-k2×BG, PG+k2×BG] are recorded as abnormal neighboring values ​​and removed from the neighboring window, where k2 is the set scaling factor. Divide the remaining adjacent windows into two equal parts in sequence, and denote them as the front window and the back window respectively. Calculate the standard deviations of the front window and the back window respectively, and denote them as VL and VR in sequence. Calculate min(VL, VR) / max(VL, VR), denoted as the fluctuation conservation coefficient SV. If SV is less than k3, then RC... j The abnormal values ​​of the cold liquid temperature difference are marked and the corresponding acquisition time is recorded as the abnormal time. Otherwise, they are marked as normal values. The process of filtering out all abnormal values ​​and their corresponding abnormal times in the cold liquid temperature difference sequence is repeated, where k3 is the set threshold.

5. The real-time intelligent control method for a motor cooling system based on edge computing according to claim 4, characterized in that, The process of performing first and second filtering on motor temperature and current data, and water cooling temperature and flow rate data to obtain current, temperature, and flow rate change data also includes the following sub-steps: For motor temperature sequence, load current sequence, coolant flow rate sequence, and coolant temperature difference sequence, remove the motor temperature, load current, coolant flow rate, and coolant temperature difference collected at any abnormal moment from the corresponding sequence. After completion, the first temperature sequence, the first current sequence, the first flow rate sequence, and the first temperature difference sequence are obtained respectively; The first temperature sequence is divided into multiple subsequences using the abnormal time as the dividing position. Any subsequence is denoted as the first subsequence. The first difference sequence corresponding to the first subsequence is calculated and denoted as the first difference sequence. Let any first difference in the first difference numerator sequence be denoted as FC. m Where m represents the position number, taken as the distance FC in the first temperature sequence. m The most recent e4 data points, where e4 is the number of settings; denoted as FC. m Reference data, obtain all reference data and FC m The median FM.

6. The real-time intelligent control method for a motor cooling system based on edge computing according to claim 5, characterized in that, The process of performing first and second filtering on motor temperature and current data, and water cooling temperature and flow rate data to obtain current, temperature, and flow rate change data also includes the following sub-steps: Calculate the absolute difference between all reference data and FM, and obtain the median CM of all absolute differences; if |FC m If -FM| is greater than k4×CM, then FC will be applied. m This is recorded as an abnormal difference, and the two corresponding acquisition times are also marked as abnormal times, where k4 is the set scaling factor; Repeatedly screen all abnormal differences and corresponding abnormal times in the first difference sequence to obtain the corresponding screening subsequence. Repeatedly obtain the screening subsequences corresponding to all subsequences to obtain the screening difference sequence corresponding to the first temperature sequence. Repeatedly obtain the filtered differential sequences corresponding to the current sequence, the first flow rate sequence, and the first temperature difference sequence, and remove all first-order differences involved in any abnormal moment from the corresponding filtered differential sequences. After completion, the differential temperature sequence, differential current sequence, differential flow rate sequence, and differential temperature difference sequence are obtained respectively, and are recorded as current, temperature and flow rate change data.

7. The real-time intelligent control method for a motor cooling system based on edge computing according to claim 6, characterized in that, Based on a multilayer sensor, and by establishing a temperature change model for the water-cooled motor using current, temperature, and flow rate change data, the process of obtaining the motor temperature change model also includes the following sub-steps: The first-order differences in the differential temperature sequence, differential current sequence, differential flow rate sequence, and differential temperature difference sequence are respectively denoted as differential temperature, differential current, differential flow rate, and differential temperature difference. An initial model is constructed based on a multilayer perceptron. The initial model includes an input layer, a hidden layer, and an output layer. The input of the initial model is set to the differential current, differential flow rate, and differential temperature difference at the same acquisition time, and the output is the differential temperature. The data on changes in current, temperature, and flow rate are divided into a training set and a test set. The initial model is trained using the training set to obtain the motor temperature change model. The motor temperature change model is then tested using the test set. The mean absolute error of the motor temperature change model is calculated and denoted as MAE.

8. The real-time intelligent control method for a motor cooling system based on edge computing according to claim 7, characterized in that, During the operation of the water-cooled motor, the corresponding current, temperature, and flow rate are collected, and accurate temperature analysis is performed based on the motor temperature change model. Real-time control of the cooling system includes the following sub-steps: During normal operation of the first motor, the motor temperature, load current, coolant temperature difference and coolant flow rate of the first motor are synchronously collected at a first time interval; and the differential temperature ET, differential current, differential flow rate and differential temperature difference corresponding to the current collection time are calculated. The differential current, differential flow rate, and differential temperature difference are input into the motor temperature change model to obtain the differential temperature output by the model, which is denoted as the theoretical temperature difference LT. If ET is not located in [k5×(LT-MAE), k5×(LT+MAE)], the motor temperature collected at the current acquisition time is marked as the noise temperature, and the flow rate of the coolant is not controlled. Here, k5 is the set proportional coefficient. If ET is located in [k5×(LT-MAE), k5×(LT+MAE)], then the motor temperature collected at the current acquisition time is marked as the accurate temperature, and the flow rate of the coolant is controlled by PID based on the current acquired motor temperature.

Images