A predictive maintenance system and method for a high voltage frequency converter

Abstract

The application relates to the field of high-voltage frequency converter operation and maintenance, and particularly discloses a high-voltage frequency converter predictive maintenance system and method, which comprises the following steps: constructing an electro-thermal coupling physical model based on the electrical topology and heat conduction structure of the high-voltage frequency converter; constructing a physical constraint error based on various physical constraints, and constructing a total loss function of a state prediction model based on a data fitting error and the physical constraint error; collecting real-time electrical parameters and thermal field parameters in the operation process of the high-voltage frequency converter, inputting the trained state prediction model, and outputting state prediction parameters of the high-voltage frequency converter; and analyzing the state change trend of key components of the high-voltage frequency converter based on the state prediction parameters, and dynamically adjusting the operation parameters of the high-voltage frequency converter. The application can reduce the prediction failure problem of the high-voltage frequency converter under small sample fault conditions and unknown conditions, and improve the stability and physical consistency of the state prediction parameters under complex conditions.

Description

Technical Field

[0001] This invention relates to the field of high-voltage frequency converter operation and maintenance technology, and more specifically to a predictive maintenance system and method for high-voltage frequency converters. Background Technology

[0002] Existing fault detection technologies for high-voltage frequency converters mainly include threshold monitoring and rule-based diagnostic methods. These methods are simple and intuitive, but they cannot analyze fault change trends and their settings rely on experience, which can easily lead to false alarms or missed alarms.

[0003] High-voltage frequency converters operate under conditions of bidirectional coupling of electrothermal parameters, small-sample fault conditions, and strong interference. Traditional data-driven prediction models are prone to errors where predicted parameters deviate from actual physical laws. Furthermore, due to the low frequency of severe faults in high-voltage frequency converters, it is difficult to obtain sample data covering all fault scenarios during actual operation. Traditional data-driven prediction models rely entirely on historical fault data for training, making them susceptible to prediction failure under unknown operating conditions.

[0004] Therefore, how to reduce the predictive failure problem of high-voltage frequency converters under small sample fault conditions and unknown conditions, and improve the stability and physical consistency of state prediction parameters under complex conditions, is an urgent problem to be solved by those skilled in the art. Summary of the Invention

[0005] In view of the above problems, the present invention proposes a predictive maintenance system and method for high-voltage frequency converters, so as to overcome the above problems or at least partially solve the above problems.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] In a first aspect, the present invention provides a predictive maintenance method for a high-voltage frequency converter, comprising the following steps: An electrothermal coupling physical model is constructed based on the electrical topology and heat conduction structure of the high-voltage frequency converter, and various physical constraints are established based on the electrothermal coupling physical model. A physical constraint error is constructed based on multiple physical constraints, and a total loss function for the state prediction model is constructed based on the data fitting error and the physical constraint error. When training the state prediction model, the weight ratio of the data fitting error and the physical constraint error in the total loss function is adaptively adjusted according to the number of historical fault samples. Real-time electrical and thermal parameters of the high-voltage frequency converter during operation are collected and input into the trained state prediction model, and the state prediction parameters of the high-voltage frequency converter are output. Based on the state prediction parameters, the state change trend of key components of the high-voltage frequency converter is analyzed, and the operating parameters of the high-voltage frequency converter are dynamically adjusted when the state change trend of a certain key component is abnormal.

[0008] Furthermore, the expression for the total loss function is:

[0009] in, and For adaptive dynamic weighting coefficients, The data fitting error is... The physical constraint error; when the number of historical fault samples exceeds the threshold, As the data fitting error increases, it becomes dominant in the total loss function; when historical fault samples are below a threshold or under unknown operating conditions... As the error increases, the physical constraint error becomes dominant in the total loss function.

[0010] Furthermore, the process of constructing the physical constraint error includes: A thermal conduction constraint error is constructed based on thermal conduction constraints, which are used to characterize the dynamic balance between heat generation and heat diffusion during the operation of a high-voltage frequency converter. A power loss constraint error is constructed based on the power loss constraint, which is used to characterize the correspondence between power loss and heat generation during the operation of power devices in high-voltage frequency converters. An electrothermal coupling constraint error is constructed based on thermoelectric coupling constraints, which is used to characterize the bidirectional feedback relationship between the thermal field parameters and electrical parameters of power devices. The physical constraint error is obtained by weighted summing of the heat conduction constraint error, the power loss constraint error, and the electrothermal coupling constraint error, and its expression is as follows:

[0011] in, The thermal conduction constraint error, The power loss constraint error is... The electrothermal coupling constraint error, , , These are the corresponding weighting coefficients.

[0012] Furthermore, the expression for the thermal conduction constraint error is:

[0013] in, For material density, For the specific heat capacity of the material, Number of thermal sampling nodes For the first Temperature of each thermal field node, For the first Each thermal field node receives a heat source input. Thermal conductivity, For the first Each thermal field node receives a heat source input.

[0014] Furthermore, the expression for the power loss constraint is:

[0015] in, The number of thermal field sampling nodes, where j is the j-th thermal field sampling node; The predicted power loss value output by the state prediction model. This represents the theoretical power loss value calculated based on power loss constraints.

[0016] Furthermore, the expression for the thermoelectric coupling constraint error is as follows:

[0017] in, The number of thermal field sampling nodes, where j is the j-th thermal field sampling node; The theoretical on-resistance is calculated based on the aforementioned thermoelectric coupling constraint formula. The predicted on-resistance is the output of the state prediction model.

[0018] Furthermore, the expression for the data fitting error is:

[0019] in, The number of thermal field sampling nodes, where j is the j-th thermal field sampling node; The predicted value of the power unit node temperature output by the state prediction model. These are the measured values ​​of the power unit node temperature. The predicted values ​​of the electrical parameters of the power unit nodes output by the state prediction model. These are the measured values ​​of the electrical parameters of the power unit nodes.

[0020] Furthermore, when the state prediction parameters show that the power device has an abnormal temperature rise trend, the operating parameters of the high-voltage frequency converter are dynamically adjusted according to the state change trend, including reducing the output power, adjusting the PWM switching frequency, and adjusting the carrier modulation strategy.

[0021] Furthermore, before adjusting the operating parameters of the high-voltage frequency converter, the adjusted operating parameters are first input into the state prediction model to predict the state change trend of the key components of the high-voltage frequency converter after the adjustment, which is used as the test result; when the test result meets the preset operating conditions, the adjusted operating parameters are then sent to the high-voltage frequency converter.

[0022] Secondly, the present invention provides a predictive maintenance system for high-voltage frequency converters, which adopts the predictive maintenance method for high-voltage frequency converters as described above, including: a data acquisition module, an electrothermal coupling physical model construction module, a model coupling training module, a prediction module, and an operation and maintenance management module; The data acquisition module is used to collect electrical and thermal parameters of the high-voltage frequency converter in real time during operation.

[0023] The electrothermal coupling physical model construction module is used to construct an electrothermal coupling physical model based on the electrical topology and heat conduction structure of the high-voltage frequency converter, and to establish physical constraint errors based on the electrothermal coupling physical model; The model coupling training module is used to construct physical constraint errors based on multiple physical constraints, and to construct the total loss function of the state prediction model based on the data fitting error and the physical constraint error. When training the state prediction model, the weight ratio of the data fitting error and the physical constraint error in the total loss function is adaptively adjusted according to the number of historical fault samples. The prediction module is used to input the real-time electrical parameters and thermal field parameters of the high-voltage frequency converter during operation into the trained state prediction model, and output the state prediction parameters of the high-voltage frequency converter. The operation and maintenance management module is used to analyze the state change trend of key components of the high-voltage frequency converter based on state prediction parameters, and dynamically adjust the operating parameters of the high-voltage frequency converter when the state change trend of a certain key component is abnormal.

[0024] As can be seen from the above technical solution, compared with the prior art, the present invention has the following beneficial effects: 1. This invention differs from rule-based diagnosis that relies solely on experience or purely data-driven models. By establishing a physical constraint mechanism for electrothermal coupling of high-voltage frequency converters, the training process of the state prediction model is constrained. When historical data is scarce or unknown and complex operating conditions are encountered, physical laws will dominate the model deduction, ensuring that the state prediction parameters still meet the preset physical law constraints under small sample fault conditions and unknown operating conditions. This ensures that the predicted trajectory does not deviate from the boundaries of physical laws, thereby improving the reliability of the state prediction parameters, effectively reducing false alarms, and mitigating prediction failures under complex operating conditions.

[0025] 2. When an anomaly is detected, this invention first analyzes the adjustment parameters based on the state prediction parameters before adjusting the operating parameters of the high-voltage frequency converter. Only when the test results are stable and meet the preset operating conditions is the operating command issued to the high-voltage frequency converter, thus ensuring the safety and reliability of the high-voltage frequency converter operation and slowing down the rate of fault deterioration. Attached Figure Description

[0026] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0027] Figure 1 This is a flowchart of the predictive maintenance method for high-voltage frequency converters provided in an embodiment of the present invention; Figure 2 This is a structural block diagram of the predictive maintenance system for high-voltage frequency converters provided in an embodiment of the present invention. Detailed Implementation

[0028] 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.

[0029] like Figure 1 As shown in the figure, an embodiment of the present invention discloses a predictive maintenance method for a high-voltage frequency converter, comprising the following steps: S1. Construct an electrothermal coupling physical model based on the electrical topology and heat conduction structure of the high-voltage frequency converter, and establish various physical constraints based on the electrothermal coupling physical model; S2. Construct physical constraint error based on multiple physical constraints, and construct the total loss function of the state prediction model based on data fitting error and physical constraint error. When training the state prediction model, adaptively adjust the weight ratio of data fitting error and physical constraint error in the total loss function according to the number of historical fault samples. S3. Collect real-time electrical and thermal parameters during the operation of the high-voltage frequency converter, input them into the trained state prediction model, and output the state prediction parameters of the high-voltage frequency converter. S4. Analyze the state change trend of key components of the high-voltage frequency converter based on state prediction parameters, and dynamically adjust the operating parameters of the high-voltage frequency converter when the state change trend of a certain key component is abnormal.

[0030] The specific implementation methods of each component of the system of the present invention will be further explained below.

[0031] S1. Construction of the electrothermal coupling physical model and physical constraints, specifically including: A circuit simulation model is established based on the electrical topology of the high-voltage frequency converter, and a thermal simulation model is established based on the device structure and heat dissipation structure. The circuit simulation model is used to obtain parameters such as conduction loss, switching loss, and power loss; the thermal simulation model is used to obtain parameters such as thermal conductivity, heat source parameters, and thermal field distribution parameters; the electrothermal coupling model is used to obtain parameters such as the temperature coefficient of conduction resistance, the temperature rise coefficient of power loss, and thermal feedback parameters. These parameters are used to construct thermal conduction constraints, power loss constraints, and electrothermal coupling constraints, respectively.

[0032] S2. Construct a state prediction model based on a neural network and use historical operating data as training samples to train the model. During training, a total loss function is constructed based on data fitting error and physical constraint error to constrain the model. The physical constraint error is constructed based on thermal conduction constraints, power loss constraints, and electrothermal coupling constraints.

[0033] The construction process of the total loss function will be further explained below.

[0034] 1) Constructing thermal conduction constraint errors based on thermal conduction constraints. Thermal conduction constraints are used to characterize the dynamic balance between heat generation and heat diffusion during the operation of high-voltage frequency converters. In actual operation, the heat generated by power devices diffuses to the surrounding structure through thermal conduction paths, thus forming changes in the thermal field distribution. By establishing thermal conduction constraints, the thermal field results output by the state prediction model are made to satisfy the laws of heat conservation and heat diffusion, so as to avoid the state prediction parameters deviating from reality.

[0035] The expression for the thermal conduction constraint is:

[0036] in, For material density, For the specific heat capacity of the material, For temperature, Thermal conductivity, These are the parameters of the heat source.

[0037] The expression for the thermal conduction constraint error based on the thermal conduction constraint is as follows:

[0038] in, Number of thermal sampling nodes For the first Temperature of each thermal field node, For the first Each thermal field node receives a heat source input.

[0039] 2) Constructing a power loss constraint error based on power loss constraints. Power loss constraints characterize the relationship between power loss and heat generation during the operation of power devices in a high-voltage frequency converter. In actual operation, the conduction and switching losses of power devices together form a heat source input and affect the device junction temperature. By establishing power loss constraints, the power loss parameters output by the state prediction model are made to meet the power variation law under actual operating conditions.

[0040] The power loss constraint expression is:

[0041] in, To conduct current, For on-resistance, For switching frequency, These are the conduction loss parameters. These are the shutdown loss parameters.

[0042] The expression for the power loss constraint error based on the power loss constraint is as follows:

[0043] in, The predicted power loss value is output by the state prediction model. This represents the theoretical power loss value calculated based on power loss constraints.

[0044] 3) An electrothermal coupling constraint error is constructed based on thermoelectric coupling constraints. This constraint characterizes the bidirectional feedback relationship between the thermal and electrical parameters of power devices. In actual operation, power loss leads to an increase in device junction temperature, which further affects the on-resistance, switching losses, and current-carrying capacity of the power devices, thus causing changes in electrical parameters. By establishing electrothermal coupling constraints, the predicted parameters output by the state prediction model are made to satisfy the coupling variation law between electrical and thermal parameters.

[0045] The expression for the electrothermal coupling constraint is:

[0046] in, For temperature The on-resistance below, The on-resistance at the reference temperature, This is a temperature reference coefficient. This is a reference temperature.

[0047] The expression for the error of electrothermal coupling constraint based on electrothermal coupling constraint is as follows:

[0048] in, The theoretical on-resistance is calculated based on electrothermal coupling constraints. This is the predicted on-resistance output by the state prediction model.

[0049] 4) Physical constraint error characterizes the degree to which the state prediction parameters output by the state prediction model satisfy the preset physical laws. The physical constraint error comprehensively considers heat conduction constraints, power loss constraints, and electrothermal coupling constraints. By measuring the deviation between the output parameters of the state prediction model and the physical constraint relationship, it limits the output of state prediction parameters that violate the preset physical laws. The physical constraint error is obtained by weighted summing of the heat conduction constraint error, power loss constraint error, and electrothermal coupling constraint error, and its expression is: , in, For thermal conduction constraint error, To constrain the error due to power loss, For electrothermal coupling constraint error, , , These are the corresponding weighting coefficients.

[0050] 5) Data fitting error is used to characterize the deviation between the output parameters of the state prediction model and the actual measured parameters, including deviations in thermal field parameters and electrical parameters. Its expression is:

[0051] in, These are predicted values ​​for the power unit node temperature. These are the measured values ​​of the power unit node temperature. These are the predicted values ​​of the electrical parameters of the power unit nodes. These are the measured values ​​of the electrical parameters of the power unit nodes.

[0052] 6) The total loss function characterizes the degree of data fit between the state prediction parameters output by the state prediction model and the actual operating data, as well as the degree of satisfaction with the preset physical laws. By simultaneously introducing data fitting error and physical constraint error, the state prediction model learns the changing patterns of historical operating data while maintaining the state prediction parameters' compliance with preset physical law constraints, thereby improving the physical consistency of the state prediction parameters under complex operating conditions and small sample conditions. Its expression is:

[0053] in, and For adaptive dynamic weighting coefficients, For data fitting error, This refers to the physical constraint error.

[0054] When there are enough historical fault samples (i.e., greater than or equal to the preset threshold), As the data fitting error increases, it becomes the dominant factor in the total loss function, and the state prediction model mainly learns the pattern of equipment state changes based on historical operating data.

[0055] When there are insufficient historical fault samples (i.e., less than the preset threshold) or the operating condition is unknown, As the physical constraint error increases, it becomes dominant in the total loss function, ensuring that the state prediction parameters continuously meet the preset physical law constraints. Ultimately, this ensures that even in the absence of fault samples, the model's state prediction parameters remain constrained within the preset physical laws, reducing the model's failure problem under unknown conditions.

[0056] S3. Collect real-time electrical and thermal parameters during the operation of the high-voltage frequency converter, input them into the trained state prediction model, and output the state prediction parameters of the high-voltage frequency converter.

[0057] S4. Analyze the state change trend of key components of the high-voltage frequency converter based on state prediction parameters, and dynamically adjust the operating parameters of the high-voltage frequency converter when the state change trend of a certain key component is abnormal.

[0058] Specifically, when the status prediction parameters show that the power device has an abnormal temperature rise trend, the module dynamically adjusts the operating parameters of the high-voltage frequency converter according to the status change trend, including reducing the output power, adjusting the PWM switching frequency, and adjusting the carrier modulation strategy.

[0059] Before adjusting the operating parameters, the adjusted parameters are input into the state prediction model. Based on the state prediction model, a virtual operation analysis is performed on the adjusted operating conditions to predict the temperature rise trend of the power devices and the system stability. When the test results meet the preset operating conditions, the corresponding operating parameter adjustment strategy is executed to reduce the thermal stress on the power devices and delay further deterioration of the fault. Finally, real-time operating data after the parameter adjustment is collected again and re-input into the state prediction model to continuously update the state prediction parameters of the high-voltage frequency converter.

[0060] In other embodiments, the present invention provides a predictive maintenance system for high-voltage frequency converters, which employs the predictive maintenance method for high-voltage frequency converters as described above, including: a data acquisition module, an electrothermal coupling physical model construction module, a model coupling training module, a prediction module, and an operation and maintenance management module; The data acquisition module is used to collect electrical and thermal parameters of the high-voltage frequency converter in real time during operation.

[0061] The electrothermal coupling physical model building module is used to build an electrothermal coupling physical model based on the electrical topology and heat conduction structure of the high-voltage frequency converter, and to establish various physical constraints based on the electrothermal coupling physical model; The model coupling training module is used to construct physical constraint errors based on multiple physical constraints, and to construct the total loss function of the state prediction model based on the data fitting error and the physical constraint error. When training the state prediction model, the weight ratio of the data fitting error and the physical constraint error in the total loss function is adaptively adjusted according to the number of historical fault samples. The prediction module is used to input the real-time electrical and thermal parameters of the high-voltage frequency converter during operation into the trained state prediction model and output the state prediction parameters of the high-voltage frequency converter. The operation and maintenance management module is used to analyze the status change trend of key components of the high-voltage frequency converter based on status prediction parameters, and dynamically adjust the operating parameters of the high-voltage frequency converter when the status change trend of a certain key component is abnormal.

[0062] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0063] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A predictive maintenance method for high-voltage frequency converters, characterized in that, Includes the following steps: An electrothermal coupling physical model is constructed based on the electrical topology and heat conduction structure of the high-voltage frequency converter, and various physical constraints are established based on the electrothermal coupling physical model. A physical constraint error is constructed based on multiple physical constraints, and a total loss function for the state prediction model is constructed based on the data fitting error and the physical constraint error. When training the state prediction model, the weight ratio of the data fitting error and the physical constraint error in the total loss function is adaptively adjusted according to the number of historical fault samples. Real-time electrical and thermal parameters of the high-voltage frequency converter during operation are collected and input into the trained state prediction model, and the state prediction parameters of the high-voltage frequency converter are output. Based on the state prediction parameters, the state change trend of key components of the high-voltage frequency converter is analyzed, and the operating parameters of the high-voltage frequency converter are dynamically adjusted when the state change trend of a certain key component is abnormal.

2. The predictive maintenance method for high-voltage frequency converters as described in claim 1, characterized in that, The expression for the total loss function is: in, and For adaptive dynamic weighting coefficients, The data fitting error is... The physical constraint error; when the number of historical fault samples exceeds the threshold, As the data fitting error increases, it becomes dominant in the total loss function; when historical fault samples are below a threshold or under unknown operating conditions... As the error increases, the physical constraint error becomes dominant in the total loss function.

3. The predictive maintenance method for high-voltage frequency converters as described in claim 2, characterized in that, The process of constructing the physical constraint error includes: A thermal conduction constraint error is constructed based on thermal conduction constraints, which are used to characterize the dynamic balance between heat generation and heat diffusion during the operation of a high-voltage frequency converter. A power loss constraint error is constructed based on the power loss constraint, which is used to characterize the correspondence between power loss and heat generation during the operation of power devices in high-voltage frequency converters. An electrothermal coupling constraint error is constructed based on thermoelectric coupling constraints, which is used to characterize the bidirectional feedback relationship between the thermal field parameters and electrical parameters of power devices. The physical constraint error is obtained by weighted summing of the heat conduction constraint error, the power loss constraint error, and the electrothermal coupling constraint error, and its expression is as follows: in, The thermal conduction constraint error, The power loss constraint error is... The electrothermal coupling constraint error, , , These are the corresponding weighting coefficients.

4. The predictive maintenance method for high-voltage frequency converters as described in claim 3, characterized in that, The expression for the thermal conduction constraint error is: in, For material density, For the specific heat capacity of the material, Number of thermal sampling nodes For the first Temperature of each thermal field node, For the first Each thermal field node receives a heat source input. Thermal conductivity, For the first Each thermal field node receives a heat source input.

5. The predictive maintenance method for high-voltage frequency converters as described in claim 3, characterized in that, The expression for the power loss constraint is: in, The number of thermal field sampling nodes, where j is the j-th thermal field sampling node; The predicted power loss value output by the state prediction model. This represents the theoretical power loss value calculated based on power loss constraints.

6. The predictive maintenance system for high-voltage frequency converters as described in claim 3, characterized in that, The expression for the thermoelectric coupling constraint error is: in, The number of thermal field sampling nodes, where j is the j-th thermal field sampling node; The theoretical on-resistance is calculated based on the aforementioned thermoelectric coupling constraint formula. The predicted on-resistance is the output of the state prediction model.

7. The predictive maintenance system for high-voltage frequency converters as described in claim 2, characterized in that, The expression for the data fitting error is: in, The number of thermal field sampling nodes, where j is the j-th thermal field sampling node; The predicted value of the power unit node temperature output by the state prediction model. These are the measured values ​​of the power unit node temperature. The predicted values ​​of the electrical parameters of the power unit nodes output by the state prediction model. These are the measured values ​​of the electrical parameters of the power unit nodes.

8. The predictive maintenance method for high-voltage frequency converters as described in claim 3, characterized in that, When the state prediction parameters show that the power device has an abnormal temperature rise trend, the operating parameters of the high-voltage frequency converter are dynamically adjusted according to the state change trend, including reducing the output power, adjusting the PWM switching frequency, and adjusting the carrier modulation strategy.

9. The predictive maintenance method for high-voltage frequency converters as described in claim 8, characterized in that, Before adjusting the operating parameters of the high-voltage frequency converter, the adjusted operating parameters are first input into the state prediction model to predict the state change trend of the key components of the high-voltage frequency converter after the adjustment, which is used as the test result. When the test results meet the preset operating conditions, the adjusted operating parameters are then sent to the high-voltage frequency converter.

10. A predictive maintenance system for a high-voltage frequency converter, characterized in that, It adopts the predictive maintenance method for high-voltage frequency converters as described in any one of claims 1-9, including: a data acquisition module, an electrothermal coupling physical model construction module, a model coupling training module, a prediction module, and an operation and maintenance management module; The data acquisition module is used to collect electrical and thermal parameters of the high-voltage frequency converter in real time during operation. The electrothermal coupling physical model construction module is used to construct an electrothermal coupling physical model based on the electrical topology and heat conduction structure of the high-voltage frequency converter, and to establish various physical constraints based on the electrothermal coupling physical model; The model coupling training module is used to construct physical constraint errors based on multiple physical constraints, and to construct the total loss function of the state prediction model based on the data fitting error and the physical constraint error. When training the state prediction model, the weight ratio of the data fitting error and the physical constraint error in the total loss function is adaptively adjusted according to the number of historical fault samples. The prediction module is used to input the real-time electrical parameters and thermal field parameters of the high-voltage frequency converter during operation into the trained state prediction model, and output the state prediction parameters of the high-voltage frequency converter. The operation and maintenance management module is used to analyze the state change trend of key components of the high-voltage frequency converter based on state prediction parameters, and dynamically adjust the operating parameters of the high-voltage frequency converter when the state change trend of a certain key component is abnormal.

Images