Method for optimizing feedback control of quantum gas sensing system of power transformation equipment

Abstract

The invention provides a method for optimizing feedback control of a quantum gas sensing system of power transformation equipment, and the method comprises the steps: determining a Gaussian mixture kernel function, and obtaining the current measurement data of the gas concentration of the power transformation equipment, the Gaussian mixture kernel function comprising the number of Gaussian kernels and the bandwidth of each Gaussian kernel; according to the current measurement data, the bandwidth of each Gaussian kernel is adjusted in the Gaussian kernel mixture function, the number of Gaussian kernels is adjusted, and the current gas concentration probability distribution is obtained; according to the current gas concentration probability distribution and a pre-constructed quantum importance function used for representing the importance degree of the current measurement data, the current Fisher information amount is calculated after the current measurement data is sampled; and adjusting the quantum gas sensing system according to the current Fisher information amount until the quantum gas sensing system is maintained in a Fisher information amount maximization state. Therefore, the design and feedback control strategy of the measurement system can be optimized, and the measurement performance of the system is further improved.

Description

Technical Field

[0001] The present application relates to the technical field of gas detection in power systems, and in particular to a method for optimizing feedback control of a quantum gas sensing system for power transformation equipment. Background Art

[0002] In modern power systems, the operating status of substation equipment is directly related to the safety, stability and reliability of the entire power grid. Gas leakage detection is one of the key links in substation equipment maintenance, especially in key equipment such as transformers and circuit breakers, where changes in gas composition are often early warning signals of equipment failure. Therefore, high-precision and high-sensitivity gas detection technology is of vital importance for preventive maintenance and fault diagnosis of power systems.

[0003] Traditional gas detection methods mainly include infrared absorption method, thermal conductivity method and chemical sensor method. The infrared absorption method has certain selectivity, but it is insufficient in low-concentration gas detection; the thermal conductivity method is simple to operate, but the detection accuracy is low and it is easily affected by changes in ambient temperature; the chemical sensor method is fast, but has poor long-term stability and is susceptible to chemical pollution. The limitations of these traditional methods make it difficult to meet the needs of modern power systems for high precision, high brightness and high dynamic range response. Recently, quantum gas sensing technology has gradually become a research hotspot in the field of gas detection due to its unique physical properties and technical advantages. It uses the high accuracy and high precision characteristics of quantum states to achieve accurate measurement of tiny gas concentration changes. Asymmetric quantum measurement technology significantly improves detection accuracy by amplifying tiny gas concentration changes, providing a new solution for gas detection in power systems.

[0004] Although quantum measurement technology has extremely high precision potential in theory, it still faces many challenges in practical applications. When encountering a mixture of multiple gases or the environment causing noise that is not high-precision, single-core calculations cannot accurately complete the probability distribution; at the same time, fixed bandwidth will lead to over-smoothing descriptions of high-concentration areas (information loss) and over-sensitivity (noise amplification) in low-concentration areas. Therefore, how to further improve the measurement performance of the system by optimizing the design of the measurement system and the feedback control strategy is still an urgent problem to be solved. Summary of the invention

[0005] The purpose of this application is to solve at least one of the above-mentioned technical deficiencies, especially the technical deficiencies in the prior art of how to further improve the measurement performance of the system by optimizing the design of the measurement system and the feedback control strategy.

[0006] In a first aspect, the present application provides a method for optimizing feedback control of a quantum gas sensing system for power conversion equipment, the method comprising:

[0007] Determine the mixture Gaussian kernel function and obtain the current measurement data of the gas concentration of the power transformation equipment. The mixture Gaussian kernel function includes Gaussian kernels and the bandwidth of each Gaussian kernel;

[0008] According to the current measurement data, adjust the bandwidth of each Gaussian kernel in the mixture Gaussian kernel function and adjust the number of Gaussian kernels to obtain the current gas concentration probability distribution;

[0009] According to the current gas concentration probability distribution and the pre-constructed quantum importance function representing the importance degree of the current measurement data, sample the current measurement data and calculate the current Fisher information;

[0010] Adjust the quantum gas sensing system according to the current Fisher information until the quantum gas sensing system maintains the state of maximizing the Fisher information.

[0011] In one embodiment, the step of adjusting the bandwidth of each Gaussian kernel in the mixture Gaussian kernel function and adjusting the number of Gaussian kernels according to the current measurement data includes:

[0012] After determining the local Fisher information of each Gaussian kernel, determine the basic bandwidth of each Gaussian kernel according to the local Fisher information of each Gaussian kernel;

[0013] Combine the basic bandwidth of each Gaussian kernel and the time variation to adjust the bandwidth of each Gaussian kernel in the mixture Gaussian kernel function;

[0014] Determine the current signal-to-noise ratio of the current measurement data and update the number of Gaussian kernels in the mixture Gaussian kernel function according to the current signal-to-noise ratio.

[0015] In one embodiment, the step of combining the basic bandwidth of each Gaussian kernel and the time variation to adjust the bandwidth of each Gaussian kernel in the mixture Gaussian kernel function includes:

[0016] Adjust the bandwidth of each Gaussian kernel in the mixture Gaussian kernel function according to the following formula:

[0017]

[0018] Among them, represents the time-varying correction term, represents the basic bandwidth, is the attenuation coefficient, is the system time constant, is the previous update time interval, represents the gas concentration parameter.

[0019] In one embodiment, the step of updating the number of Gaussian kernels in the mixture Gaussian kernel function according to the current signal-to-noise ratio includes:

[0020] Update the number of Gaussian kernels in the mixture Gaussian kernel function according to the following formula:

[0021]

[0022] where, represents the updated number of kernels at time , represents the signal-to-noise ratio, represents the gas concentration parameter at time t.

[0023] In one embodiment, the step of calculating the current Fisher information amount after sampling the current measurement data according to the current gas concentration probability distribution and the pre-constructed quantum importance function for representing the importance degree of the current measurement data includes:

[0024] Determine the current high-information amount region and the current low-information amount region in the current gas concentration probability distribution according to the current gas concentration probability distribution and the pre-constructed quantum importance function for representing the importance degree of the current measurement data;

[0025] Sample the current measurement data in the current high-information amount region by quantum projection sampling method and in the current low-information amount region by MCMC sampling method to obtain the current sampling samples;

[0026] Determine the current Fisher information amount by integral calculation according to the current sampling samples and the quantum importance function.

[0027] In one embodiment, the step of sampling the current measurement data in the current high-information amount region by quantum projection sampling method and in the current low-information amount region by MCMC sampling method includes:

[0028] Sample the current measurement data in the current high-information amount region by quantum projection sampling method according to the following formula:

[0029]

[0030] Sample the current measurement data in the current low-information amount region by MCMC sampling method according to the following formula:

[0031]

[0032] where, is the quantum projection measurement operator, represents the quantum state related to the gas concentration parameter g, represents the quantum importance function for measuring the importance degree of the data point x.

[0033] In one embodiment, the step of determining the current Fisher information by integral calculation according to the current sampling sample and the quantum importance function includes:

[0034] Calculate the current Fisher information according to the following formula:

[0035]

[0036] where, represents the current number of samples, and , represents the standard deviation of the current integral estimate, represents the integral value, represents the current gas concentration probability distribution, represents the measurement data point the probability density at which it appears, represents the gas concentration parameter, represents the value of the quantum importance function at the sample point at.

[0037] In one embodiment, the step of adjusting the quantum gas sensing system according to the current Fisher information includes:

[0038] According to the current Fisher information, the feedback gain parameter of the quantum gas sensing system is dynamically adjusted by using the gradient ascent algorithm, and the iteration step size is adaptively optimized by the learning rate.

[0039] In a second aspect, the present application provides a device for optimizing the feedback control of a quantum gas sensing system for substation equipment. The device includes:

[0040] A mixed Gaussian kernel function determination module, configured to determine a mixed Gaussian kernel function and obtain current measurement data of the gas concentration of the substation equipment. The mixed Gaussian kernel function includes Gaussian kernels and the bandwidth of each Gaussian kernel;

[0041] A gas concentration probability distribution determination module, configured to adjust the bandwidth of each Gaussian kernel and the number of Gaussian kernels in the mixed Gaussian kernel function according to the current measurement data to obtain the current gas concentration probability distribution;

[0042] A Fisher information calculation module, configured to calculate the current Fisher information after sampling the current measurement data according to the current gas concentration probability distribution and a pre-constructed quantum importance function representing the importance degree of the current measurement data;

[0043] A quantum gas sensing system adjustment module, configured to adjust the quantum gas sensing system according to the current Fisher information until the quantum gas sensing system maintains the state of maximizing the Fisher information.

[0044] In a third aspect, the present application provides a computer device, including: one or more processors, and a memory;

[0045] The memory stores computer-readable instructions, which, when executed by the one or more processors, perform the steps of the method for optimizing feedback control of the quantum gas sensing system for substation equipment as described in any one of the above embodiments.

[0046] As can be seen from the above technical solutions, the embodiments of the present application have the following advantages:

[0047] The method for optimizing feedback control of the quantum gas sensing system for substation equipment provided by the present application solves the deficiencies of single-core computing under complex gas mixtures and noise interference through dynamic adjustment of the mixture Gaussian kernel function, can more accurately describe the gas concentration probability distribution, and avoids the problems of information loss and noise amplification caused by a fixed bandwidth. Sampling measurement data according to importance using the quantum importance function focuses on key data, improves the calculation efficiency and the rationality of resource allocation. Taking the Fisher information as the optimization target, the sensing system is adjusted in real-time feedback to ensure that it is always in the optimal measurement state, thereby enhancing the sensitivity and adaptability of the system to gas concentration changes, improving the measurement accuracy and stability, and providing more reliable data support for the operation monitoring of substation equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0049] Figure 1 It is a schematic flowchart of the method for optimizing feedback control of the quantum gas sensing system for substation equipment provided by the embodiments of the present application;

[0050] Figure 2 It is a schematic structural diagram of the device for optimizing feedback control of the quantum gas sensing system for substation equipment provided by the embodiments of the present application;

[0051] Figure 3 It is a schematic internal structure diagram of the computer device provided by the embodiments of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0052] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present application.

[0053] The present application provides a method for optimizing feedback control of a quantum gas sensing system for substation equipment. Among them, substation equipment is an important equipment in the power system for voltage level conversion, power distribution, and circuit control. It is mainly applied in places such as substations and is responsible for transmitting electrical energy from the transmission line to the distribution network or the user side. Its functions include boosting, stepping down, power distribution, circuit protection, and control, etc. It is a key component to ensure the safe, stable, and efficient operation of the power system; the quantum gas sensing system is a system that uses quantum technology to measure gas concentration and has the characteristics of high precision and high sensitivity, and is suitable for gas monitoring in complex environments such as substation equipment. The following embodiments will be described by taking the application of this method to a computer device as an example. It can be understood that the computer device can be various devices with data processing functions, including but not limited to a single server, a server cluster, a personal laptop, a desktop computer, etc. As Figure 1 shown, the method may include the following steps:

[0054] S101: Determine the mixture Gaussian kernel function and obtain the current measurement data of the gas concentration of the substation equipment. The mixture Gaussian kernel function includes Gaussian kernels and the bandwidth of each Gaussian kernel.

[0055] Among them, the mixture Gaussian kernel function is a probability density function linearly combined by multiple Gaussian kernel functions and is used to describe the data distribution. Each Gaussian kernel function has its corresponding mean and covariance matrix. By adjusting these parameters, a better fit to the complex data distribution can be achieved. The number of Gaussian kernels refers to the number of Gaussian kernels included in the mixture Gaussian kernel function, that is, the number of mixture components. Different numbers of Gaussian kernels will affect the fitting degree of the model to the data distribution. In the Gaussian kernel function, the bandwidth determines the width of the Gaussian distribution, that is, the influence range of data points under this kernel function. A smaller bandwidth will make the model more sensitive to local data, while a larger bandwidth will make the model more concerned about the overall data trend. The current measurement data of the gas concentration of the substation equipment refers to the concentration values of the gas around the substation equipment obtained in real time through sensors and other devices, and these data reflect the gas leakage situation under the operating state of the substation equipment.

[0056] In this step, by analyzing the historical data and prior knowledge of the gas concentration of the substation equipment, appropriate Gaussian kernel numbers and initial bandwidths are selected to construct a mixture Gaussian kernel function. For example, the EM algorithm can be used to estimate the parameters of the mixture Gaussian model, including the mean, covariance matrix, and mixing coefficients of each Gaussian kernel, so as to determine a mixture Gaussian kernel function that can better describe the gas concentration distribution. Then, by connecting to the gas sensor, the real-time gas concentration value transmitted by the sensor can be received. When acquiring data, factors such as the accuracy of the sensor, sampling frequency, and data transmission stability need to be considered to ensure that the acquired data can accurately reflect the true concentration of the gas around the substation equipment.

[0057] It can be understood that in the measurement of the gas concentration of substation equipment, determining the mixture Gaussian kernel function and acquiring the current measurement data can achieve accurate measurement and effective monitoring. By reasonably selecting and adjusting the mixture Gaussian kernel function, the concentration distribution in a complex gas mixture environment can be more accurately described, avoiding the information loss and noise amplification problems caused by fixed bandwidth and single-core calculation. At the same time, obtaining accurate measurement data in real time provides a reliable data basis for subsequent probability distribution calculation and system optimization, helps to improve the accuracy and stability of the entire measurement system, and provides a strong guarantee for the safe operation of substation equipment.

[0058] S102: According to the current measurement data, adjust the bandwidth of each Gaussian kernel and the number of Gaussian kernels in the mixture Gaussian kernel function to obtain the current gas concentration probability distribution.

[0059] Among them, the current gas concentration probability distribution is the distribution of the probabilities of the gas concentration appearing in different regions obtained by using probability density estimation methods such as the mixture Gaussian kernel function based on the current measurement data, reflecting the possible values of the gas concentration and their corresponding probabilities.

[0060] In this step, by analyzing the current measurement data, algorithms such as the expectation maximization algorithm EM are used to estimate the parameters of the mixture Gaussian model, including the mean, covariance matrix, and mixing coefficients of each Gaussian kernel. According to the distribution characteristics of the data, the bandwidth of each Gaussian kernel is dynamically adjusted. For example, the bandwidth is appropriately reduced in the area where data points are dense to capture local details, and the bandwidth is appropriately increased in the area where data points are sparse to cover the overall trend. At the same time, according to the complexity and distribution of the data, through model selection methods such as the Bayesian information criterion BIC or the Akaike information criterion AIC, the optimal number of Gaussian kernels is determined to avoid overfitting or underfitting of the model. For example, in the measurement of the gas concentration of substation equipment, when it is detected that the gas concentration change is relatively complex and there are multiple peaks, the number of Gaussian kernels is increased to more accurately describe this multi-peak distribution; while when the gas concentration change is relatively stable, the number of Gaussian kernels is reduced to simplify the model and improve the calculation efficiency.

[0061] After adjusting the parameters of the mixture Gaussian kernel function, use this function to perform probability density estimation on the current measurement data, calculate the probability that each data point belongs to this mixture distribution, so as to obtain the probability distribution of the entire gas concentration. This can be achieved by multiplying the density function values of each Gaussian kernel by the corresponding mixing coefficients and summing them up. Finally, a continuous probability distribution curve that can reflect the probability of gas concentration appearing in different regions is obtained.

[0062] It can be understood that in the measurement of the gas concentration of substation equipment, adjusting the bandwidth and the number of Gaussian kernels of the mixture Gaussian kernel function can achieve accurate measurement and effective monitoring. By reasonably adjusting the bandwidth of each Gaussian kernel, it can more flexibly adapt to the gas concentration change characteristics in different regions, avoiding the information loss and noise amplification problems caused by a fixed bandwidth, and thus more accurately describe the probability distribution of the gas concentration. At the same time, dynamically adjusting the number of Gaussian kernels according to the actual situation of the data can make the model complexity match the data characteristics, neither overfitting the noise due to too many kernels nor failing to capture the true data distribution pattern due to too few kernels, improving the generalization ability and measurement accuracy of the model. In addition, this dynamic adjustment mechanism also enhances the adaptability of the system to complex environments, enabling it to better cope with various gas leakage situations that may occur during the operation of substation equipment, and providing more reliable data support for the safe operation of the equipment.

[0063] S103: According to the current gas concentration probability distribution and the pre-constructed quantum importance function used to represent the importance degree of the current measurement data, sample the current measurement data and then calculate the current Fisher information.

[0064] Among them, the quantum importance function is a pre-constructed function used to measure the importance degree of each data point in the current measurement data. In the quantum gas sensing system, this function assigns weights to different data points according to the characteristics of the data, such as the distribution density of the data points, the correlation with the equipment operation state, etc., to guide the subsequent sampling process and make the calculation more focused on key data. The Fisher information is an index that measures the amount of information about unknown parameters contained in the observed data. In the parameter estimation problem, the larger the Fisher information, the higher the estimation accuracy of the observed data for the unknown parameters, that is, the smaller the variance of the estimated value. Its mathematical definition is the variance of the first derivative of the log-likelihood function, or equivalently, the negative expectation of the second derivative of the log-likelihood function.

[0065] In this step, first, based on the current gas concentration probability distribution, the distribution characteristics of the data points are determined. Then, in combination with the quantum importance function, which assigns a weight value to each data point, and the magnitude of the weight value reflects the importance of the data point in the measurement. According to these weights, a weighted sampling method, such as weighted random sampling, is used to select a certain number of samples from the current measurement data. For example, in the gas concentration monitoring of substation equipment, if the gas concentration changes in certain areas have a greater impact on the operation status of the equipment, the quantum importance function will assign higher weights to the data points in these areas, and the probability of these data points being selected during sampling is greater, thus ensuring that subsequent calculations pay more attention to the data in key areas.

[0066] After obtaining the sampled data, calculations are performed according to the definition of Fisher information. For example, first, construct the likelihood function of the observed data, which describes the probability of observing these data under given parameters; then, take the logarithm of the likelihood function to obtain the log-likelihood function; next, take the first and second derivatives of the log-likelihood function respectively to obtain the Score function and its second derivative; finally, according to the formula of Fisher information, calculate the variance of the Score function or the expectation of the negative value of the second derivative of the log-likelihood function. In actual calculations, numerical methods or statistical software packages can be used to approximately calculate these derivatives and expectations. For example, for complex probability distribution models, the Monte Carlo method can be used for numerical integration to estimate the value of Fisher information.

[0067] It can be understood that in the gas concentration measurement of substation equipment, sampling the measurement data according to the current gas concentration probability distribution and the quantum importance function and calculating the Fisher information are key steps to achieve system optimization and high-precision measurement. Guided by the quantum importance function for sampling can focus on key data points, improve the utilization efficiency of computing resources, and ensure fine analysis of sensitive areas of gas concentration changes. Calculating the Fisher information provides a quantitative basis for system optimization, enabling the system to dynamically adjust measurement parameters and strategies according to the magnitude of the information, thereby improving measurement accuracy and stability, enhancing the system's adaptability to complex environments, and providing more reliable data support for the safe operation of substation equipment.

[0068] S104: Adjust the quantum gas sensing system according to the current Fisher information until the quantum gas sensing system maintains the state of maximizing the Fisher information.

[0069] In this step, first, obtain the currently calculated Fisher information value. Then, this value can be compared with a preset threshold or target value, and through a feedback control algorithm, such as the proportional-integral-derivative control algorithm PID, etc., relevant parameters of the quantum gas sensing system can be adjusted. For example, when the current Fisher information is lower than the target value, the system may increase the sampling frequency of the sensor, adjust the sensitivity of the sensor, or change the focusing degree of the measurement area, etc., to increase the information content. In the gas concentration monitoring of power transformation equipment, if it is detected that the gas concentration change in a certain area contributes significantly to the Fisher information, the system will automatically adjust the measurement focus of the sensor and strengthen the monitoring of this area. Through continuous feedback adjustment, the measurement parameters of the system are continuously monitored and optimized, so that the Fisher information of the system gradually increases and finally stabilizes near the maximum value. When the Fisher information of the system reaches the maximum value, it means that the current measurement data has the highest estimation accuracy for the gas concentration parameter, and the system is in the optimal measurement state. At this time, the system will maintain this state to ensure the accurate monitoring of the gas concentration of power transformation equipment.

[0070] It can be understood that in the gas concentration measurement of power transformation equipment, by adjusting the quantum gas sensing system according to the current Fisher information until it maintains the state of maximizing the Fisher information, high-precision and high-stability measurement can be achieved. Through this adjustment, the system can dynamically optimize the measurement parameters, ensure the accurate capture and efficient utilization of key data, thereby improving the sensitivity and response speed to gas concentration changes. This not only enhances the adaptability of the system in a complex environment but also provides more reliable and accurate data support for the safe operation of power transformation equipment, effectively improving the performance and reliability of the entire monitoring system.

[0071] In the above embodiment, through the dynamic adjustment of the mixed Gaussian kernel function, the deficiencies of single-core calculation under complex gas mixtures and noise interference are solved, and the gas concentration probability distribution can be described more accurately, avoiding the problems of information loss and noise amplification caused by a fixed bandwidth. Using the quantum importance function to sample the measurement data according to the importance degree, focusing on key data, and improving the calculation efficiency and the rationality of resource allocation. Taking the Fisher information as the optimization goal, the sensing system is adjusted in real-time by feedback to ensure that it is always in the optimal measurement state, thereby enhancing the sensitivity and adaptability of the system to gas concentration changes, improving the measurement accuracy and stability, and providing more reliable data support for the operation monitoring of power transformation equipment.

[0072] In one embodiment, the steps of adjusting the bandwidth of each Gaussian kernel and the number of Gaussian kernels in the mixed Gaussian kernel function according to the current measurement data include:

[0073] After determining the local Fisher information of each Gaussian kernel, determine the basic bandwidth of each Gaussian kernel according to the local Fisher information of each Gaussian kernel;

[0074] Adjust the bandwidth of each Gaussian kernel in the mixture Gaussian kernel function by combining the basic bandwidth of each Gaussian kernel and the time variation.

[0075] Determine the current signal-to-noise ratio of the current measurement data, and update the number of Gaussian kernels in the mixture Gaussian kernel function according to the current signal-to-noise ratio.

[0076] Among them, the local Fisher information is the Fisher information calculated based on the current measurement data on each Gaussian kernel in the mixture Gaussian kernel function, and is used to measure the amount of information about the gas concentration parameter contained in the data region corresponding to the Gaussian kernel. The initial bandwidth determined according to the local Fisher information of each Gaussian kernel is used to ensure that the main features of the data region described by the corresponding Gaussian kernel can be accurately captured at this bandwidth. The time variation refers to the change in the data distribution and characteristics over time during the measurement process due to factors such as the environment and equipment status. The signal-to-noise ratio is the ratio of the signal to the noise, and is used to measure the strength of the effective signal in the measurement data relative to the noise, and is an important indicator for evaluating the data quality and measurement accuracy.

[0077] Specifically, first calculate the local Fisher information of each Gaussian kernel based on the current measurement data. Then, according to the magnitude of this information, determine the basic bandwidth of each Gaussian kernel according to a certain rule or formula. For example, in the gas concentration monitoring of substation equipment, if the gas concentration in the area corresponding to a certain Gaussian kernel changes rapidly, its local Fisher information is large, and at this time, a smaller basic bandwidth is allocated to it to more accurately capture the characteristics of the rapidly changing signal; on the contrary, if the gas concentration changes slowly, a larger basic bandwidth is allocated to cover a wider data range.

[0078] After obtaining the basic bandwidth of each Gaussian kernel, consider the impact of time variation on the data distribution. For example, as the monitoring time progresses, the gas leakage pattern around the substation equipment may change, resulting in more drastic changes in the gas concentration in some areas, while other areas tend to be stable. By monitoring these changes in real time, dynamically adjust the bandwidth of each Gaussian kernel. For areas with drastic changes, appropriately reduce the bandwidth to improve the ability to capture local details; for areas with stable changes, appropriately increase the bandwidth to cover a wider data range and improve the calculation efficiency.

[0079] After that, the signal-to-noise ratio of the current measurement data can be calculated by estimating the ratio of the signal power to the noise power. Then, it is determined whether to update the number of Gaussian kernels according to the magnitude of the signal-to-noise ratio. In the case of high signal-to-noise ratio, the effective signals in the data dominate, and the number of Gaussian kernels can be appropriately increased to describe the data distribution more meticulously; while in the case of low signal-to-noise ratio, the noise has a greater impact on the data, and too many Gaussian kernels may lead to overfitting of the model to the noise. At this time, the number of Gaussian kernels should be reduced to simplify the model and improve its robustness to the true signals. For example, in the gas concentration monitoring of substation equipment, when it is detected that the environmental noise suddenly increases and causes the signal-to-noise ratio to drop, the system automatically reduces the number of Gaussian kernels to avoid the model being interfered by the noise and ensure the effective monitoring of the gas concentration.

[0080] In one example, the mixture Gaussian kernel function can be expressed as:

[0081]

[0082] where, represents the Gaussian distribution; is a function of the gas concentration g and represents the mean of the k-th kernel; is the weight coefficient of the k-th kernel; is the adaptive bandwidth of the k-th kernel.

[0083]

[0084] where, is the local Fisher information corresponding to the k-th sub-kernel:

[0085]

[0086] Determination of the quantum state-dependent mean:

[0087]

[0088] where, is the quantum projection measurement operator, satisfying , represents the quantum state related to the gas concentration parameter g.

[0089]

[0090]

[0091] where, represents the Hilbert space of the system, represents the Hamiltonian related to the gas concentration, is the initial quantum state.

[0092] In this embodiment, by determining the local Fisher information of each Gaussian kernel and accordingly determining the base bandwidth, it is possible to ensure that each Gaussian kernel accurately captures the key features of the corresponding data region and avoid information loss. Adjusting the bandwidth in combination with the time variation enables the system to dynamically adapt to changes in the environment and data distribution, improving the flexibility and accuracy of measurement. Updating the number of Gaussian kernels according to the current signal-to-noise ratio effectively balances the relationship between model complexity and data quality, enhancing the adaptability and robustness of the system under different noise conditions. In this way, the description of the probability distribution of gas concentration by the mixture Gaussian kernel function is optimized, improving the measurement accuracy and stability of the quantum gas sensing system and providing more reliable data support for the safe operation of substation equipment.

[0093] In one embodiment, the step of adjusting the bandwidth of each Gaussian kernel in the mixture Gaussian kernel function in combination with the base bandwidth of each Gaussian kernel and the time variation includes:

[0094] Adjust the bandwidth of each Gaussian kernel in the mixture Gaussian kernel function according to the following formula:

[0095]

[0096] Wherein, represents the time-varying correction term, represents the base bandwidth, is the attenuation coefficient, is the system time constant, is the previous update time interval, represents the gas concentration parameter.

[0097] Specifically, the base bandwidth can be expressed as:

[0098]

[0099] Wherein, is the reference bandwidth, determined according to the Shannon sampling theorem:

[0100]

[0101] In the measurement of the gas concentration of substation equipment, this formula can be used to dynamically adjust the bandwidth of the mixture Gaussian kernel function to adapt to the change of gas concentration over time. For example, when the gas concentration changes slowly, the bandwidth can gradually increase to cover a wider data range; while when the gas concentration changes rapidly, the bandwidth can remain small to capture the rapidly changing details.

[0102] In this embodiment, by adjusting the bandwidth, the adaptability of the system to changes in the data distribution can be improved, ensuring that the probability distribution of gas concentration can be accurately described at different time scales. This helps to improve the measurement accuracy and stability, especially in complex and dynamic environments.

[0103] In one embodiment, the step of updating the number of Gaussian kernels in the Gaussian mixture kernel function according to the current signal-to-noise ratio includes:

[0104] Update the number of Gaussian kernels in the Gaussian mixture kernel function according to the following formula:

[0105]

[0106] where, represents the updated number of kernels at time , represents the signal-to-noise ratio, represents the gas concentration parameter at time t.

[0107] In this embodiment, this formula dynamically adjusts the number of Gaussian kernels in the Gaussian mixture kernel function through the current signal-to-noise ratio. Its significance lies in adaptively optimizing the model complexity according to the data quality. In the case of a high signal-to-noise ratio, the formula will calculate a larger number of Gaussian kernels, enabling the model to describe the data distribution more meticulously; while in the case of a low signal-to-noise ratio, the formula will calculate a smaller number of Gaussian kernels to avoid overfitting of the model to noise. By updating the number of Gaussian kernels, the relationship between the model complexity and data quality can be effectively balanced, enhancing the adaptability and robustness of the system under different noise conditions. This helps to improve the accuracy and stability of the measurement, especially in complex and dynamic environments.

[0108] In one embodiment, the step of calculating the current Fisher information amount after sampling the current measurement data according to the current gas concentration probability distribution and the pre-constructed quantum importance function for representing the importance degree of the current measurement data includes:

[0109] Determine the current high-information region and the current low-information region in the current gas concentration probability distribution according to the current gas concentration probability distribution and the pre-constructed quantum importance function for representing the importance degree of the current measurement data;

[0110] Sample the current measurement data in the current high-information region by quantum projection sampling and in the current low-information region by MCMC sampling to obtain the current sampling samples;

[0111] Determine the current Fisher information amount through integral calculation according to the current sampling samples and the quantum importance function.

[0112] Among them, quantum projection sampling is a sampling method based on quantum states, which uses quantum projection measurement operators to efficiently sample data in high-information regions. MCMC sampling, that is, Markov chain Monte Carlo sampling, is an algorithm for sampling from a specific probability distribution by constructing a Markov chain to generate samples that conform to the target distribution.

[0113] Specifically, first, according to the current gas concentration probability distribution and the pre-constructed quantum importance function, the importance of each data point is evaluated. High-information regions refer to those regions that contribute more to the estimation of gas concentration parameters, usually having a higher data density or being closely related to the operating state of the device. Low-information regions, on the contrary, have fewer data points or less contribution to parameter estimation. In high-information regions, the quantum projection sampling method is used to efficiently generate sampling samples according to the weights of the quantum importance function. In low-information regions, the MCMC sampling method is adopted to gradually generate samples that conform to the target distribution by constructing a Markov chain. For example, in the gas concentration monitoring of substation equipment, high-information regions may be concentrated near the key operating parameters of the equipment, while low-information regions are distributed at the edges or in regions with high noise. Using the current sampling samples and the quantum importance function, calculate the contribution of each sample point to the Fisher information. Summarize these contributions through an integration method to obtain the current Fisher information. This can be achieved through numerical integration or statistical software packages.

[0114] In one example, the quantum importance function can be expressed as:

[0115]

[0116] Where is the quantum enhancement factor, , is the photon number operator, is the reference photon number.

[0117] In this embodiment, by determining the high-information regions and low-information regions in the current gas concentration probability distribution and respectively using the quantum projection sampling and MCMC sampling methods for sampling, key data can be efficiently obtained, and the utilization efficiency of computing resources can be improved. Quantum projection sampling quickly captures key features in high-information regions, while MCMC sampling ensures the comprehensiveness of data coverage in low-information regions. Combining the quantum importance function and integral calculation to determine the Fisher information provides a quantitative basis for system optimization, further improving the measurement accuracy and stability, and enhancing the adaptability and robustness of the system in complex environments.

[0118] In one embodiment, the step of sampling the current measurement data by quantum projection sampling in the current high-information region and by MCMC sampling in the current low-information region includes:

[0119] Sampling the current measurement data by quantum projection sampling in the current high-information region according to the following formula:

[0120]

[0121] Sampling the current measurement data by MCMC sampling in the current low-information region according to the following formula:

[0122]

[0123] where is the quantum projection measurement operator, represents the quantum state related to the gas concentration parameter g, represents the quantum importance function, which is used to measure the importance degree of the data point x.

[0124] In this embodiment, sampling in the high-information region by quantum projection sampling can efficiently capture key features by using the quantum state and the projection measurement operator, and quickly generate high-information samples. The MCMC sampling formula, on the other hand, gradually generates samples that conform to the target distribution by constructing a Markov chain in the low-information region, ensuring comprehensive data coverage. This strategy of combining the two sampling methods not only improves the utilization efficiency of computing resources, but also further enhances the measurement accuracy and stability, and strengthens the adaptability and robustness of the system in complex environments.

[0125] In one embodiment, the step of determining the current Fisher information by integral calculation according to the current sampling sample and the quantum importance function includes:

[0126] Calculating the current Fisher information according to the following formula:

[0127]

[0128] where represents the current number of samples, and , represents the standard deviation of the current integral estimate, represents the integral value, represents the current gas concentration probability distribution, represents the measurement data point the probability density of the occurrence of, represents the gas concentration parameter, represents the quantum importance function at the sample point at the value.

[0129] In this embodiment, the calculation formula of the Fisher information comprehensively evaluates the information contribution of the current measurement data to the gas concentration parameter by combining the probability density of the sample points, the derivative of the log-likelihood function, and the sample weights. Dynamically adjusting the number of samples adaptively optimizes the number of samples according to the currently estimated standard deviation and Fisher information, ensuring that under limited computing resources, both the estimation accuracy and the computing efficiency can be guaranteed. In a complex and dynamic environment, dynamic adjustment can effectively balance the relationship between computing resources and measurement accuracy, enhance the adaptability and robustness of the system, and thus improve the performance and reliability of the entire measurement system.

[0130] In one embodiment, the steps of adjusting the quantum gas sensing system according to the current Fisher information include:

[0131] According to the current Fisher information, the gradient ascent algorithm is used to dynamically adjust the feedback gain parameter of the quantum gas sensing system, and the iteration step size is adaptively optimized through the learning rate.

[0132] Among them, the gradient ascent algorithm is used to maximize the objective function, and the parameters are iteratively adjusted to increase the value of the objective function. The feedback gain parameter is a parameter used to adjust the system feedback strength, which affects the response speed and stability of the system. Adaptive learning rate optimization means dynamically adjusting the learning rate according to the current gradient information to optimize the iteration step size and improve the convergence speed and stability of the algorithm.

[0133] Specifically, first calculate the current Fisher information, and then use the gradient ascent algorithm to adjust the feedback gain parameter according to the Fisher information. For example, after calculating the gradient of the Fisher information with respect to the feedback gain parameter, update the feedback gain parameter according to the gradient direction and the learning rate to maximize the Fisher information until the preset convergence condition or the maximum number of iterations is reached.

[0134] In the gradient ascent algorithm, the magnitude of the learning rate directly affects the speed and stability of parameter update. Through the adaptive learning rate optimization method, the learning rate is dynamically adjusted according to the historical gradient information to ensure that appropriate step sizes can be used for parameter update at different stages. For example, set an initial learning rate, and in each iteration, accumulate the square of the gradient. According to the sum of the accumulated gradient squares and the initial learning rate, dynamically adjust the learning rate, and use the adjusted learning rate for parameter update.

[0135] In one example, the gradient ascent method is used to optimize the feedback control strategy:

[0136]

[0137] Among them, is the learning rate, which is used to control the step size of optimization.

[0138] In this embodiment, by dynamically adjusting the feedback gain parameter of the quantum gas sensing system using the gradient ascent algorithm according to the current Fisher information, it is possible to ensure that the system is always in the optimal measurement state, improving the accuracy and stability of the measurement. At the same time, by adaptively optimizing the iteration step size with the learning rate, it is possible to automatically adjust the step size of parameter update according to the current gradient information, avoiding oscillations caused by too large a learning rate or slow convergence caused by too small a learning rate, thereby accelerating the convergence speed of the algorithm and improving the response speed and adaptability of the system. These steps work together to optimize the performance of the system and enhance its robustness and reliability in complex environments.

[0139] To facilitate the understanding of the solution of this application, specific examples are provided below for illustration.

[0140] The following uses the method provided by the present invention for optimizing the feedback control of the quantum gas sensing system for substation equipment to estimate the gas absorption coefficient parameter, uses the non-orthogonal quantum measurement technology with increased feedback control to amplify the measurement of the small absorption coefficient of the system, and adjusts the system by calculating the Fisher information, and finally obtains a higher-precision estimated value of the gas absorption coefficient.

[0141] (1) Experimental equipment:

[0142] 1. Use a 1550 nm band ultra-narrow linewidth laser (NKT Photonics Koheras BASIK), with a linewidth ≤ 1 kHz and a power stability of ±0.1%, to generate an initial quantum state through a fiber polarization controller:

[0143]

[0144] 2. Use a gas absorption cell with a cavity length L = 10 cm, L = 10 cm, and an anti-reflection coating (reflectivity < 0.01%) on the inner wall. The temperature control accuracy of this equipment is ±0.01 °C (using a PID temperature control module), and an electric field sensor is built-in (range 0 - 100 kV / m, accuracy ±0.5 kV / m);

[0145] 3. Use a 16-channel InGaAs APD array (Princeton Lightwave PGA-080), with a single-photon detection efficiency ≥ 35%, a dark count rate ≤ 200 cps, and a time resolution of 50 ps.

[0146] 4. Use an Xilinx Zynq UltraScale+ FPGA chip, which integrates the following core modules: a kernel density estimation accelerator (throughput 2.4 T ops / s); a quantum Monte Carlo coprocessor (supporting a hybrid sampling strategy); an adaptive control logic unit (response delay ≤ 5 μs).

[0147] (2) Model initialization:

[0148]

[0149] Initial number of cores , constraint , .

[0150] Initial bandwidth reference calculation:

[0151]

[0152] (3) Dynamic parameter update

[0153] Calculate the local Fisher information:

[0154]

[0155] Update the weights at this time:

[0156]

[0157] Real variable bandwidth correction:

[0158]

[0159] Core number adjustment:

[0160] When the SNR changes :

[0161] Core fission: For the largest core, insert two new cores.

[0162] Core merger: For the smallest core, execute .

[0163] (4) Quantum-classical hybrid integral calculation

[0164] Quantum measurement projection:

[0165] Implement state projection through a Mach-Zehnder interferometer (switching time < 10 ns), sampling efficiency ≥ 100 k samples / s.

[0166] In the low-gradient region, adopt:

[0167] .

[0168] Real-time error estimation:

[0169]

[0170] Adaptive adjustment of the number of samples:

[0171]

[0172] It can be found from the expression of the Fisher information quantity that the value is related to and the value of is related to the quantum amplification factor Therefore, the that maximizes the Fisher information quantity can be obtained by taking the derivative of I. Through calculation, it is obtained that when the quantum amplification factor takes the value of:

[0173]

[0174] the Fisher information quantity reaches the maximum value. The Fisher information quantity at this time can reach the quantum Fisher information quantity.

[0175] According to the calculation result of the Fisher information quantity, the gradient ascent method is adopted to gradually optimize the feedback control parameters and dynamically adjust the feedback control strategy to maximize the Fisher information quantity:

[0176]

[0177] where is the learning rate, which is used to control the step size of optimization. After each adjustment, the Fisher information quantity is recalculated until it reaches the maximum value or meets the preset optimization goal.

[0178] Next, the device for optimizing the feedback control of the quantum gas sensing system for substation equipment provided by the embodiments of the present application will be described. The device for optimizing the feedback control of the quantum gas sensing system for substation equipment described below can be mutually corresponding and referred to the method for optimizing the feedback control of the quantum gas sensing system for substation equipment described above. As Figure 2 shown, the present application provides a device for optimizing the feedback control of the quantum gas sensing system for substation equipment. The device includes:

[0179] A mixed Gaussian kernel function determination module 201, configured to determine a mixed Gaussian kernel function and obtain current measurement data of the gas concentration of the substation equipment. The mixed Gaussian kernel function includes Gaussian kernels and the bandwidth of each Gaussian kernel;

[0180] A gas concentration probability distribution determination module 202, configured to adjust the bandwidth of each Gaussian kernel and the number of Gaussian kernels in the mixed Gaussian kernel function according to the current measurement data to obtain the current gas concentration probability distribution;

[0181] A Fisher information quantity calculation module 203, configured to calculate the current Fisher information quantity after sampling the current measurement data according to the current gas concentration probability distribution and a quantum importance function pre-constructed to represent the importance degree of the current measurement data;

[0182] The quantum gas sensing system adjustment module 204 is used to adjust the quantum gas sensing system according to the current Fisher information content until the quantum gas sensing system is maintained in the state of maximizing the Fisher information content.

[0183] In one embodiment, the gas concentration probability distribution determination module 202 includes:

[0184] The basic bandwidth determination unit is used to determine the local Fisher information content of each Gaussian kernel, and then determine the basic bandwidth of each Gaussian kernel according to the local Fisher information content of each Gaussian kernel;

[0185] The bandwidth adjustment unit is used to adjust the bandwidth of each Gaussian kernel in the mixture Gaussian kernel function by combining the basic bandwidth of each Gaussian kernel and the time variation;

[0186] The Gaussian kernel number update unit is used to determine the current signal-to-noise ratio of the current measurement data and update the number of Gaussian kernels in the mixture Gaussian kernel function according to the current signal-to-noise ratio.

[0187] In one embodiment, the bandwidth adjustment unit includes:

[0188] The bandwidth adjustment subunit is used to adjust the bandwidth of each Gaussian kernel in the mixture Gaussian kernel function according to the following formula:

[0189]

[0190] Where represents the time-varying correction term, represents the basic bandwidth, is the attenuation coefficient, is the system time constant, is the last update time interval, represents the gas concentration parameter.

[0191] In one embodiment, the Gaussian kernel number update unit includes:

[0192] The Gaussian kernel number update subunit is used to update the number of Gaussian kernels in the mixture Gaussian kernel function according to the following formula:

[0193]

[0194] Where represents the updated number of kernels at time , represents the signal-to-noise ratio, represents the gas concentration parameter at time t.

[0195] In one embodiment, the Fisher information content calculation module 203 includes:

[0196] An information amount region determination subunit, configured to determine a current high-information amount region and a current low-information amount region in the current gas concentration probability distribution according to the current gas concentration probability distribution and a quantum importance function pre-constructed for representing the importance degree of current measurement data;

[0197] A sample sampling unit, configured to sample the current measurement data in the current high-information amount region by means of quantum projection sampling and in the current low-information amount region by means of MCMC sampling to obtain a current sampling sample;

[0198] A Fisher information amount calculation unit, configured to determine a current Fisher information amount by integral calculation according to the current sampling sample and the quantum importance function.

[0199] In one embodiment, the sample sampling unit includes:

[0200] A first sample sampling subunit, configured to sample the current measurement data in the current high-information amount region by means of quantum projection sampling according to the following formula:

[0201]

[0202] A second sample sampling subunit, configured to sample the current measurement data in the current low-information amount region by means of MCMC sampling according to the following formula:

[0203]

[0204] Wherein, is a quantum projection measurement operator, represents a quantum state related to the gas concentration parameter g, represents the quantum importance function for measuring the importance degree of a data point x.

[0205] In one embodiment, the Fisher information amount calculation unit includes:

[0206] A Fisher information amount calculation subunit, configured to calculate the current Fisher information amount according to the following formula:

[0207]

[0208] Wherein, represents the current number of samples, and , represents the standard deviation of the current integral estimate, represents the integral value, represents the current gas concentration probability distribution, represents a measurement data point the probability density at which it appears, Represents the gas concentration parameter, represents the value of the quantum importance function at the sample point location.

[0209] In one embodiment, the quantum gas sensing system adjustment module 204 includes:

[0210] A quantum gas sensing system adjustment unit, configured to dynamically adjust the feedback gain parameter of the quantum gas sensing system according to the current Fisher information amount, and adaptively optimize the iteration step size through the learning rate.

[0211] In one embodiment, the present application further provides a storage medium, in which computer-readable instructions are stored. When the computer-readable instructions are executed by one or more processors, the one or more processors are caused to execute the steps of the method for optimizing the feedback control of the quantum gas sensing system for substation equipment as described in any one of the above embodiments.

[0212] In one embodiment, the present application further provides a computer device, in which computer-readable instructions are stored. When the computer-readable instructions are executed by one or more processors, the one or more processors are caused to execute the steps of the method for optimizing the feedback control of the quantum gas sensing system for substation equipment as described in any one of the above embodiments.

[0213] Schematically, as Figure 3 shown, Figure 3 is an internal structure schematic diagram of a computer device provided by an embodiment of the present application. The computer device 300 can be provided as a server. Referring to Figure 3 , the computer device 300 includes a processing component 302, which further includes one or more processors, and memory resources represented by a memory 301 for storing instructions executable by the processing component 302, such as application programs. The application programs stored in the memory 301 may include one or more modules each corresponding to a set of instructions. In addition, the processing component 302 is configured to execute instructions to perform the method for optimizing the feedback control of the quantum gas sensing system for substation equipment in any of the above embodiments.

[0214] The computer device 300 may further include a power supply component 303 configured to perform power management of the computer device 300, a wired or wireless network interface 304 configured to connect the computer device 300 to a network, and an input / output (I / O) interface 305. The computer device 300 can operate based on an operating system stored in the memory 301, such as WindowsServer TM, Mac OS XTM, Unix TM, Linux TM, Free BSDTM or the like.

[0215] Those skilled in the art can understand that Figure 3 the structure shown in Figure 3 is only a block diagram of some structures related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine some components, or have different component arrangements.

[0216] Finally, it should also be noted that in this text, relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprising", "including" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "comprising one..." does not exclude the presence of additional identical elements in the process, method, article or device comprising the element. In this text, "a", "an", "the", "this" and "its" may also include the plural form unless the context clearly indicates otherwise. "Plurality" means at least two cases, such as 2, 3, 5 or 8, etc. "And / or" includes any and all combinations of the related listed items.

[0217] The various embodiments in this specification are described in a progressive manner. Each embodiment focuses on the differences from other embodiments. The various embodiments can be combined as needed, and the same or similar parts can be referred to each other.

[0218] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be obvious to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to these embodiments shown herein, but will be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for optimizing feedback control of a quantum gas sensing system for power transformation equipment, characterized in that: The method comprises: Determine a mixed Gaussian kernel function, and obtain current measurement data of gas concentration of the substation equipment, wherein the mixed Gaussian kernel function includes the number of Gaussian kernels and the bandwidth of each Gaussian kernel; According to the current measurement data, adjusting the bandwidth of each Gaussian kernel in the mixed Gaussian kernel function and adjusting the number of Gaussian kernels to obtain the current gas concentration probability distribution; According to the current gas concentration probability distribution and the pre-constructed quantum importance function used to represent the importance of the current measurement data, the current Fisher information is calculated after sampling the current measurement data; The quantum gas sensing system is adjusted according to the current Fisher information amount until the quantum gas sensing system is maintained in a Fisher information maximization state.

2. The method for optimizing feedback control of a quantum gas sensing system for power transformation equipment according to claim 1, characterized in that: The step of adjusting the bandwidth of each Gaussian kernel in the mixed Gaussian kernel function and adjusting the number of Gaussian kernels according to the current measurement data comprises: After determining the local Fisher information amount of each Gaussian kernel, determining the basic bandwidth of each Gaussian kernel according to the local Fisher information amount of each Gaussian kernel; In combination with the basic bandwidth and time variation of each Gaussian kernel, adjusting the bandwidth of each Gaussian kernel in the mixed Gaussian kernel function; A current signal-to-noise ratio of the current measurement data is determined, and the number of Gaussian kernels in the mixed Gaussian kernel function is updated according to the current signal-to-noise ratio.

3. The method for optimizing feedback control of a quantum gas sensing system for power transformation equipment according to claim 2, characterized in that: The step of adjusting the bandwidth of each Gaussian kernel in the mixed Gaussian kernel function in combination with the basic bandwidth and time variation of each Gaussian kernel comprises: The bandwidth of each Gaussian kernel in the mixed Gaussian kernel function is adjusted according to the following formula: ； in, represents the time-varying correction term, Indicates the basic bandwidth, is the attenuation coefficient, is the system time constant, is the last update interval, Represents the gas concentration parameter.

4. The method for optimizing feedback control of a quantum gas sensing system for power transformation equipment according to claim 2, characterized in that: The step of updating the number of Gaussian kernels in the mixed Gaussian kernel function according to the current signal-to-noise ratio comprises: The number of Gaussian kernels in the mixed Gaussian kernel function is updated according to the following formula: ； in, Indicates at time The updated number of cores is represents the signal-to-noise ratio, Represents the gas concentration parameter at time t.

5. The method for optimizing feedback control of a quantum gas sensing system for power transformation equipment according to claim 1, characterized in that: The step of calculating the current Fisher information amount after sampling the current measurement data according to the current gas concentration probability distribution and the pre-constructed quantum importance function for indicating the importance of the current measurement data comprises: Determine a current high information content region and a current low information content region in the current gas concentration probability distribution according to the current gas concentration probability distribution and a pre-constructed quantum importance function for indicating the importance of the current measurement data; The current measurement data is sampled by quantum projection sampling in the current high information content region and by MCMC sampling in the current low information content region to obtain the current sampling sample; According to the current sample and the quantum importance function, the current Fisher information is determined by integral calculation.

6. The method for optimizing feedback control of a quantum gas sensing system for power transformation equipment according to claim 5, characterized in that: The step of sampling the current measurement data by quantum projection sampling in the current high information content region and by MCMC sampling in the current low information content region comprises: The current measurement data is sampled in the current high information content area by quantum projection sampling according to the following formula: ； The current measurement data is sampled in the current low information area by MCMC sampling according to the following formula: ； in, is the quantum projection measurement operator, represents the quantum state associated with the gas concentration parameter g, Represents the quantum importance function, which is used to measure the importance of the data point x.

7. The method for optimizing feedback control of a quantum gas sensing system for power conversion equipment according to claim 5, characterized in that: The step of determining the current Fisher information amount by integral calculation according to the current sample and the quantum importance function comprises: The current Fisher information is calculated according to the following formula: ； in, represents the current number of samples, and , represents the standard deviation of the current integral estimate, represents the integral value, represents the current gas concentration probability distribution, Represents a measurement data point The probability density of occurrence, represents the gas concentration parameter, Represents the quantum importance function at the sample point The value at .

8. The method for optimizing feedback control of a quantum gas sensing system for power transformation equipment according to any one of claims 1 to 7, characterized in that: The step of adjusting the quantum gas sensing system according to the current Fisher information amount comprises: According to the current Fisher information, the feedback gain parameter of the quantum gas sensing system is dynamically adjusted by using a gradient ascent algorithm, and the iteration step size is adaptively optimized by using a learning rate.

9. A device for optimizing feedback control of quantum gas sensing system of power substation equipment, characterized in that: The device comprises: A mixed Gaussian kernel function determination module, used to determine a mixed Gaussian kernel function and obtain current measurement data of gas concentration of the substation equipment, wherein the mixed Gaussian kernel function includes the number of Gaussian kernels and the bandwidth of each Gaussian kernel; A gas concentration probability distribution determination module, configured to adjust the bandwidth of each Gaussian kernel in the mixed Gaussian kernel function and the number of Gaussian kernels according to current measurement data, so as to obtain a current gas concentration probability distribution; A Fisher information calculation module is used to calculate the current Fisher information after sampling the current measurement data according to the current gas concentration probability distribution and a pre-built quantum importance function used to represent the importance of the current measurement data; The quantum gas sensing system adjustment module is used to adjust the quantum gas sensing system according to the current Fisher information amount until the quantum gas sensing system is maintained in a Fisher information maximization state.

10. A computer device, characterized in that: include: one or more processors, and memory; The memory stores computer-readable instructions, which, when executed by the one or more processors, perform the steps of the method for optimizing feedback control of a quantum gas sensing system for power conversion equipment as claimed in any one of claims 1 to 8.

Images