Robustness evaluation method for distribution transformer on-line detection system
By establishing internal and external parameters models, evaluating the robustness of the online detection system of the distribution transformer, the problem of lack of robustness evaluation methods in the prior art is solved, and the stability and reliability optimization of the system are achieved.
Patent Information
- Application Number
- CN202410631118.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-05-21
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2044-05-21
AI Technical Summary
The prior art lacks effective methods to evaluate the robustness of the online detection system of distribution transformers, which affects the stability and reliability of the system.
By establishing internal reference models and external reference models, a robust quantitative model is constructed, and the robustness of the online detection system of the distribution transformer is evaluated. Combining the influencing factors and external environmental factors in the system, the calculation performance robustness, dynamic adaptability, and fault recovery are combined with the influencing factors in the system and external environmental factors. Indicators such as capability and environmental interaction complexity.
The quantifiable evaluation of the robustness of the distribution transformer online detection system is achieved, which makes up for the gaps in the prior art and helps to optimize the performance of the detection system.
Smart Images

Figure CN119004749A_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the technical field of transformer parameter detection, and in particular relates to a robustness evaluation method for an online detection system of a distribution transformer. Background Art
[0002] With the development of my country's economy, electricity consumption has increased, and users' requirements for power supply reliability have continued to increase. Transformers are important power transmission and distribution equipment in the power system, especially 35kV and below distribution transformers, which are widely distributed and large in number. These distribution transformers directly face end users, and their performance directly affects the safe and stable operation of the power system. Therefore, ensuring its safe, reliable and economical operation is a vital issue. Realizing non-invasive online detection of key parameters of distribution transformers provides a reliable reference basis for its operation in the best working state, which is conducive to the stable and economical operation of the power system. On the other hand, it also provides great convenience for the transformer detection process, and has good practicality and economy.
[0003] Taking online detection as an example, it is to measure the relevant parameters of the transformer without power outage. It mainly collects the primary and secondary voltage and current data of the distribution transformer during normal operation online through voltage and current transformers, and then processes these real-time data accordingly to calculate the relevant parameters of the transformer, including short-circuit impedance, no-load current, no-load loss, load loss, reactive power loss and rated capacity.
[0004] Since online detection needs to be implemented through a system structure composed of acquisition modules, processors, etc., and this online detection system is an external system, there are many influencing factors in its operation, including not only the detection stability of the internal system, but also the stable state or aging state of the detection system in the external environment. Abnormal changes in these factors may affect the robustness of the detection system. However, there are few recorded methods for evaluating the robustness of online detection systems for distribution transformers. Summary of the invention
[0005] In view of the above technical problems, this application proposes a robustness evaluation method for an online detection system of a distribution transformer. The specific technical solution is as follows:
[0006] The robustness evaluation method of the online detection system of the distribution transformer includes the following steps:
[0007] (1) Establish internal and external parameter models based on the online detection system of distribution transformers;
[0008] (2) Analyze the influence characteristics of the internal reference model and the external reference model on the online detection system and further construct a robustness quantification model based on the internal reference model and the external reference model;
[0009] (3) Based on the constructed robustness quantification model, a distribution transformer online detection system robustness evaluation model is constructed by varying the internal parameter model and the external parameter model to evaluate the robustness of the distribution transformer online detection system.
[0010] As a further explanation of the above method of the present application, the process of establishing the internal reference model is as follows:
[0011] Determine the input quantity N (X1, X2, X3...X n ), model complexity C, response efficiency τ and system gain K;
[0012] Based on the input quantity N (X1, X2, X3...X n ), the complexity C of the detection model, the response efficiency τ and the system gain K, and propose an internal parameter model for explaining the online detection system of distribution transformers:
[0013]
[0014] In the above formula, F0 is the initial performance of the system; B j is the weight coefficient of the internal factor j on the system; ΔF j (t) is the deviation of internal factor j, η0 is the initial performance level, and k is the hardware aging rate constant;
[0015]
[0016] Among them, C=N+1; the response efficiency τ reflects the response efficiency of the system. The larger its value is, the faster the system processing speed and the more timely the response; the system gain K indicates the system's ability to amplify input changes. For the detection system, high gain often means better detection sensitivity; ω1+ω2+ω3=1.
[0017] As a further illustration of the above method of the present application, the external reference model is expressed as:
[0018]
[0019] In the above formula, η(t) is the initial efficiency of the detection system;
[0020] E i (t) is a function of environmental factors, which changes with time t, including temperature, humidity, and electromagnetic interference intensity; A i is the weight coefficient of the impact of environmental factor i on the detection system, reflecting the importance of the factor; α is the sensitivity coefficient of the impact of environmental factors on the detection system, representing the degree of impact of a 1-unit change in each environmental factor on system performance;
[0021] Ei,0 is the ideal or benchmark value of environmental factor i;
[0022] |E i (t)-E i,0 | represents the deviation of environmental factor i;
[0023] D(t)=η0e -kt , represents the hardware degradation model over time, η0 is the initial performance level, and k is the hardware aging rate constant, which determines the speed at which the hardware performance decreases.
[0024] As a further illustration of the above method of the present application, the robustness quantization model is:
[0025] R(t)=β1R perf (t)+β2R adapt (t)+β3R recover (t)+β4R envinteract (t);
[0026] Among them, R perf (t) is the performance robustness component, which measures the system's ability to maintain basic functions in the face of internal and external changes; R adapt (t) is the dynamic adaptability component, which measures the system's ability to automatically adjust parameters or algorithms to adapt to environmental changes; R recover (t) refers to the fault recovery capability component, which describes the ability of the system to quickly return to normal working state after experiencing a fault; R envinteract (t) reflects the complexity of environmental interaction, taking into account the interaction efficiency between the system and the changing environment; β1, β2, β3, and β4 are weight coefficients.
[0027] As a further illustration of the above method of this application, η ref is a reference threshold of system performance, R perf (t) is used to measure the tolerance of performance deviation. The closer the value of this component is to 1, the smaller the performance deviation of the system is under changes in internal and external factors, and the stronger its robustness is.
[0028] As a further illustration of the above method of this application,
[0029] I pref (t) and F perf (t) represents the evaluation value of system performance under the current environmental factors and internal parameter settings. If the system can achieve better performance through adaptive adjustment in a short time, the value of this component is high, indicating strong adaptability.
[0030] As a further illustration of the above method of this application, T detectis the fault detection time, T recover is the mean recovery time, χ is a positive adjustment coefficient, T ref is the expected reference time. The closer this component value is to 1, the faster the system fault detection and recovery is and the better the robustness is.
[0031] As a further illustration of the above method of this application, RMSE env is the root mean square of the prediction error of environmental factors, RU is the utilization rate of system resources, and the larger the value, the more efficient the system can interact with the environment and the more reasonable the resource utilization, and the stronger the robustness.
[0032] As a further explanation of the above method of the present application, based on the constructed robustness quantification model, the method for constructing a robustness evaluation model of the distribution transformer online detection system by variation of the internal reference model and the external reference model is:
[0033] Set the extreme value R max and R min ;
[0034] Select any variable P in the internal reference model or the external reference model, continuously vary from large to small or from small to large, record and calculate ΔU P , ΔU p =U max-P -U min-P , U max-P is the variable P corresponding to R in the process of variation max or R min The value of U min-P is another corresponding value;
[0035] Then the robustness evaluation model of the online detection system is:
[0036]
[0037] In the above formula, Z is the total number of variables P that can undergo continuous variation in the internal reference model and the external reference model.
[0038] The beneficial effects of the present invention are as follows: the present application combines the influencing factors within the distribution transformer online detection system and the influencing factors of the external environment in which the system is working, and constructs a robustness evaluation model to perform a quantifiable evaluation of the robustness of the distribution transformer online detection system, which is conducive to filling the gap in the existing technology for the robustness evaluation method of the electric transformer online detection system, and is also conducive to optimizing the electric transformer online detection system based on the evaluation method. BRIEF DESCRIPTION OF THE DRAWINGS
[0039] Figure 1 Shown is a flow chart of a robustness evaluation method for an online detection system of a distribution transformer in the present application. DETAILED DESCRIPTION
[0040] In order to make the purpose, technical solution and advantages of the embodiments of the present invention more clear, the technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments.
[0041] Example
[0042] Figure 1 The flowchart of the robustness evaluation method of the online detection system of the distribution transformer in the present application is shown, which includes the following steps:
[0043] (1) Establish internal and external parameter models based on the online detection system of distribution transformers;
[0044] (2) Analyze the influence characteristics of the internal reference model and the external reference model on the online detection system and further construct a robustness quantification model based on the internal reference model and the external reference model;
[0045] (3) Based on the constructed robustness quantification model, a distribution transformer online detection system robustness evaluation model is constructed by varying the internal parameter model and the external parameter model to evaluate the robustness of the distribution transformer online detection system.
[0046] In the steps of the above-mentioned evaluation method, the online detection system of the distribution transformer is based on non-invasive detection to achieve the purpose of detecting the parameter performance of the distribution transformer. It mainly uses sensors and advanced signal processing technology to monitor the electromagnetic characteristics of the transformer without destroying or affecting the normal operation of the transformer, and then calculates parameters such as short-circuit impedance.
[0047] The internal reference model is an evaluation model of the online detection system of the distribution transformer. The establishment process of the internal reference model is as follows:
[0048] Determine the input quantity N (X1, X2, X3...X n ), model complexity C, response efficiency τ and system gain K;
[0049] Based on the input quantity N (X1, X2, X3...X n ), the complexity C of the detection model, the response efficiency τ and the system gain K, and propose an internal parameter model for explaining the online detection system of distribution transformers:
[0050]
[0051] Among them, C=N+1; the response efficiency τ reflects the response efficiency of the system. The larger its value is, the faster the system processing speed and the more timely the response; the system gain K indicates the system's ability to amplify input changes. For the detection system, high gain often means better detection sensitivity; ω1+ω2+ω3=1.
[0052] The external parameter model is the influence model of the online detection system of the distribution transformer. The external parameter model can be expressed as:
[0053]
[0054] In the above formula, η(t) is the initial efficiency of the detection system;
[0055] E i (t) is a function of environmental factors, which changes with time t, including temperature, humidity, and electromagnetic interference intensity; A i is the weight coefficient of the impact of environmental factor i on the detection system, reflecting the importance of the factor; α is the sensitivity coefficient of the impact of environmental factors on the detection system, representing the degree of impact of a 1-unit change in each environmental factor on system performance;
[0056] E i,0 is the ideal or benchmark value of environmental factor i;
[0057] |E i (t)-E i,0 | represents the deviation of environmental factor i;
[0058] D(t)=η0e -kt , represents the hardware degradation model over time, η0 is the initial performance level, and k is the hardware aging rate constant, which determines the speed at which the hardware performance decreases;
[0059] The first part η(t) represents the initial detection capability of the system; the second part It represents the total impact of environmental factors on the performance of the detection system. The impact of each environmental factor is the weighted sum of the product of its deviation and the corresponding sensitivity coefficient. The third part - D(t) represents the aging effect of hardware over time.
[0060] For the internal reference model, it usually involves the internal structure of the detection system, component characteristics and algorithm processing, such as sensor accuracy, stability of the signal processing algorithm, etc., which can be expressed in the form of:
[0061]
[0062] F0 is the initial performance of the system; B j is the weight coefficient of the internal factor j on the system; ΔF j(t) is the deviation of internal factor j, η0 is the initial performance level, and k is the hardware aging rate constant.
[0063] Combining the internal reference model and the external reference model, a robust quantization model is provided:
[0064] R(t)=β1R perf (t)+β2R adapt (t)+β3R recover (t)+β4R envinteract (t);
[0065] Among them, R perf (t) is the performance robustness component, which measures the system's ability to maintain basic functions in the face of internal and external changes. η ref is a reference threshold of system performance, R perf (t) is used to measure the tolerance of performance deviation. The closer the value of this component is to 1, the smaller the performance deviation of the system is under changes in internal and external factors, and the stronger its robustness is.
[0066] R adapt (t) is the dynamic adaptability component, which measures the system's ability to automatically adjust parameters or algorithms to adapt to environmental changes. I pref (t) and F perf (t) represents the evaluation value of system performance under the current environmental factors and internal parameter settings. If the system can achieve better performance through adaptive adjustment in a short time, the value of this component is high, indicating strong adaptability.
[0067] R recover (t) refers to the fault recovery capability component, which describes the ability of the system to quickly return to normal working state after experiencing a fault. T detect is the fault detection time, T recover is the mean recovery time, χ is a positive adjustment coefficient, T ref is the expected reference time. The closer this component value is to 1, the faster the system fault detection and recovery is and the better the robustness is.
[0068] R envinteract (t) To reflect the complexity of environmental interaction, the interaction efficiency between the system and the changing environment is considered. RMSE env is the root mean square of the prediction error of environmental factors, and RU is the utilization rate of system resources. The larger the value, the more efficient the system can interact with the environment, the more reasonable the resource utilization, and the stronger the robustness.
[0069] The comprehensive robustness index R(t) is obtained by weighted summation of these four components. β1, β2, β3, and β4 are weight coefficients, which are adjusted according to system characteristics and application scenarios to reflect the importance of different aspects of robustness.
[0070] Based on the constructed robustness quantification model, the method of constructing the robustness evaluation model of the distribution transformer online detection system through the variation of the internal parameter model and the external parameter model is as follows:
[0071] Set the extreme value R max and R min ;
[0072] Select any variable P in the internal reference model or the external reference model, continuously vary from large to small or from small to large, record and calculate ΔU P , ΔU p =U max-P -U min-P , U max-P is the variable P corresponding to R during the variation process max or R min The value of U min-P is another corresponding value;
[0073] Then the robustness evaluation model of the online detection system is:
[0074]
[0075] In the above formula, Z is the total number of variables P that can undergo continuous variation in the internal reference model and the external reference model. For a distribution transformer online detection system, the larger the value of R', the more robust the system is. It should be noted that the internal reference model is the evaluation model of the distribution transformer online detection system. Some variables P are fixed by the system and therefore will not vary, such as the input quantity N (X1, X2, X3...X n ), model complexity C.
[0076] The weight coefficients recorded in this embodiment are all preset values, that is, constants, and their values can be determined specifically according to the scenario of model application.
[0077] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them.
Claims
1. A robustness evaluation method for an online detection system of a distribution transformer, characterized in that: The following steps are involved: (1) Establish internal and external parameter models based on the online detection system of distribution transformers; (2) Analyze the influence characteristics of the internal reference model and the external reference model on the online detection system and further construct a robustness quantification model based on the internal reference model and the external reference model; (3) Based on the constructed robustness quantification model, a distribution transformer online detection system robustness evaluation model is constructed by varying the internal parameter model and the external parameter model to evaluate the robustness of the distribution transformer online detection system.
2. The robustness evaluation method for online detection system of distribution transformer according to claim 1 is characterized in that: The process of establishing the internal reference model is as follows: Determine the input quantity N (X1, X2, X3...X n ), model complexity C, response efficiency τ and system gain K; Based on the input quantity N (X1, X2, X3...X n ), the complexity C of the detection model, the response efficiency τ and the system gain K, and propose an internal parameter model for explaining the online detection system of distribution transformers: In the above formula, F0 is the initial performance of the system; B j is the weight coefficient of the internal factor j on the system; ΔF j (t) is the deviation of internal factor j, η0 is the initial performance level, and k is the hardware aging rate constant; Among them, C=N+1; the response efficiency τ reflects the response efficiency of the system. The larger its value is, the faster the system processing speed and the more timely the response; the system gain K indicates the system's ability to amplify input changes. For the detection system, high gain often means better detection sensitivity; ω1+ω2+ω3=1.
3. The robustness evaluation method for online detection system of distribution transformer according to claim 2 is characterized in that: The external parameter model is expressed as: In the above formula, η(t) is the initial efficiency of the detection system; E i (t) is a function of environmental factors, which changes with time t, including temperature, humidity, and electromagnetic interference intensity; A i is the weight coefficient of the impact of environmental factor i on the detection system, reflecting the importance of the factor; α is the sensitivity coefficient of the impact of environmental factors on the detection system, representing the degree of impact of a 1-unit change in each environmental factor on system performance; E i,0 is the ideal or benchmark value of environmental factor i; |E i (t)-E i,0 | represents the deviation of environmental factor i; D(t)=η0e -kt , represents the hardware degradation model over time, η0 is the initial performance level, and k is the hardware aging rate constant, which determines the speed at which the hardware performance decreases.
4. The robustness evaluation method for online detection system of distribution transformer according to claim 3 is characterized in that: The robustness quantization model is: R(t)=β1R perf (t)+β2R adapt (t)+β3R recover (t)+β4R envinteract (t); Among them, R perf (t) is the performance robustness component, which measures the system's ability to maintain basic functions in the face of internal and external changes; R adapt (t) is the dynamic adaptability component, which measures the system's ability to automatically adjust parameters or algorithms to adapt to environmental changes; R recover (t) refers to the fault recovery capability component, which describes the ability of the system to quickly return to normal working state after experiencing a fault; R envinteract (t) reflects the complexity of environmental interaction, taking into account the interaction efficiency between the system and the changing environment; β1, β2, β3, and β4 are weight coefficients.
5. The robustness evaluation method for online detection system of distribution transformer according to claim 4 is characterized in that: η ref is a reference threshold of system performance, R perf (t) is used to measure the tolerance of performance deviation. The closer the value of this component is to 1, the smaller the performance deviation of the system is under changes in internal and external factors, and the stronger its robustness is.
6. The robustness evaluation method for online detection system of distribution transformer according to claim 4 is characterized in that: I pref (t) and F perf (t) represents the evaluation value of system performance under the current environmental factors and internal parameter settings. If the system can achieve better performance through adaptive adjustment in a short time, the value of this component is high, indicating strong adaptability.
7. The robustness evaluation method for online detection system of distribution transformer according to claim 4 is characterized in that , T detect is the fault detection time, T recover is the mean recovery time, χ is a positive adjustment coefficient, T ref is the expected reference time. The closer this component value is to 1, the faster the system fault detection and recovery is and the better the robustness is.
8. The robustness evaluation method for online detection system of distribution transformer according to claim 4 is characterized in that: RMSE env is the root mean square of the prediction error of environmental factors, RU is the utilization rate of system resources, and the larger the value, the more efficient the system can interact with the environment and the more reasonable the resource utilization, and the stronger the robustness.
9. The robustness evaluation method for online detection system of distribution transformer according to claim 4, characterized in that: Based on the constructed robustness quantification model, the method of constructing the robustness evaluation model of the distribution transformer online detection system through the variation of the internal parameter model and the external parameter model is as follows: Set the extreme value R max and R min ; Select any variable P in the internal reference model or the external reference model, continuously vary from large to small or from small to large, record and calculate ΔU P , ΔU p =U max-P -U min-P , U max-P is the variable P corresponding to R in the process of variation max or R min The value of U min-P is another corresponding value; Then the robustness evaluation model of the online detection system is: In the above formula, Z is the total number of variables P that can undergo continuous variation in the internal reference model and the external reference model.
Citation Information
Patent Citations
Robust estimation method based on synchronous phasor measurement for voltage power sensitivity of distribution network
CN107577870A
Robust state estimation method for power grid voltage sag
CN114429049A
Method and system for evaluating robustness of power algorithm model in confrontation scene
CN115564709A
AI-based distribution network 10KV line technology loss reduction intelligent adjustment method and device
CN117913987A
Machine learning model evaluation method based on signal identification
CN117972370A
