A method for automatically locating partial discharge signals

Abstract

The present invention relates to the technical field of measuring electrical variables, and specifically to a method for automatically locating a partial discharge signal, comprising the following steps: according to the initial position area of ​​the partial discharge signal source, generating multiple groups of position information randomly distributed in the initial position area of ​​the partial discharge signal source, assigning the position information to each particle, calculating the initial weight of each particle, assigning a corresponding weight to each particle, and obtaining an initial weighted particle set. The present invention achieves a wider detection range and higher positioning flexibility, generates position information randomly distributed in the initial position area of ​​the signal source, assigns an initial weight to each particle, and allows exploration in many possible discharge points. After combining multiple sensor data, the extracted signal strength and arrival time are used to update the weight of the particle, so that the weight of each particle more accurately reflects its positioning possibility.

Description

Technical Field

[0001] The invention relates to the technical field of measuring electrical variables, and in particular to a method for automatically locating a partial discharge signal. Background Art

[0002] The field of measuring electrical variables involves a wide range of technologies and methods for detecting, recording and analyzing electrical parameters such as voltage, current, power and frequency in various electrical systems and equipment. The technology in this field is not only essential for the operation of power systems, but also plays a key role in ensuring the safety and efficiency of power facilities. With the development of smart grids, the technology of measuring electrical variables is increasingly used in real-time monitoring, fault diagnosis and energy management. The automatic location method of partial discharge signals is a method for detecting and locating partial discharge signals in electrical equipment.

[0003] Existing technologies rely on a single sensor for signal analysis, which is easily affected by environmental noise and interference. In addition, when the signal strengths of the signal sources are similar or the arrival time is close, the existing technology is difficult to effectively distinguish, resulting in insufficient positioning accuracy. In the complex environment of the power system, this positioning error may lead to incorrect maintenance decisions, increasing operational risks and maintenance costs. Summary of the invention

[0004] The purpose of the present invention is to solve the shortcomings in the prior art and to propose a method for automatically locating a partial discharge signal.

[0005] In order to achieve the above object, the present invention adopts the following technical solution, a method for automatically locating a partial discharge signal, comprising the following steps:

[0006] According to the initial position area of ​​the partial discharge signal source, multiple groups of position information randomly distributed in the initial position area of ​​the partial discharge signal source are generated, the position information is assigned to each particle, the initial weight of each particle is calculated, and the corresponding weight is assigned to each particle to obtain an initial weighted particle set;

[0007] Based on the initial weighted particle set, receiving signal data from multiple sensors, extracting signal strength values ​​and arrival time values, comparing the predicted signal strength and arrival time of each particle position with the actual values, calculating the likelihood value of each particle, and updating the particle weight by multiplying the original weight by the likelihood value to obtain an updated weighted particle set;

[0008] Based on the updated weighted particle set, the cumulative distribution function of the particle weight is calculated to generate a random number sequence between zero and one. For each random number, the particles whose cumulative distribution function is greater than the random number are captured, the particles are selected to form a new particle set, and the selected particles are given equal weights to obtain a resampled particle set.

[0009] Based on the resampled particle set, the position information of all particles is multiplied by the corresponding weights, the products are accumulated to obtain the sum of weighted position information, the weights of the particles are accumulated to obtain the sum of weights, and the sum of weighted position information is divided by the sum of weights to obtain an estimated value of the signal source position.

[0010] Preferably, the steps of obtaining the initial weighted particle set are:

[0011] According to the initial position area of ​​the partial discharge signal source, multiple sets of randomly distributed position information are generated, and each particle is assigned a position to obtain a position information set;

[0012] Based on the position information set, the initial weight is calculated for each particle, and the calculation formula is:

[0013] ;

[0014] in, is the weight of the ith particle, is the estimated distance from the particle to the signal source, is the expected distance offset, is the adjustment coefficient;

[0015] Based on the calculated weights, a corresponding weight is assigned to each particle, and an initial weighted particle set is generated by combining the weights and position information.

[0016] Preferably, the predicted signal strength and arrival time of each particle position are compared with the actual value, and the specific steps of calculating the likelihood value of each particle are:

[0017] Receive signal data from multiple sensors, extract signal strength values ​​and arrival time values, and obtain initial signal data;

[0018] Based on the initial signal data, the predicted signal strength and arrival time of each particle position are compared with the actual values, and the likelihood value of each particle is calculated using the following formula:

[0019] ;

[0020] in, is the likelihood value of the ith particle, and are the predicted signal strength and arrival time, respectively. and is the observed signal strength and arrival time, is the standard deviation.

[0021] Preferably, the step of acquiring the updated weighted particle set is:

[0022] Based on the calculated likelihood value of each particle, the current weight of each particle is obtained to obtain the current weight set;

[0023] Based on the current weight set, multiply the current weight of each particle by the corresponding likelihood value, update the weight of the particle, and obtain an updated weight;

[0024] Based on the updated weight, the position information of each particle is recombined with the updated weight to generate an updated weighted particle set.

[0025] Preferably, the steps of calculating the cumulative distribution function of the particle weight and generating a random number sequence between zero and one are:

[0026] Based on the updated weighted particle set, obtaining an updated weight of each particle to obtain an updated weight set;

[0027] Based on the updated weight set, calculate the cumulative distribution function of the particle weight , the calculation formula is:

[0028] ;

[0029] in, is the updated weight of the ith particle, is the position of the particle, is the target location, is the standard deviation, n is the total number of particles;

[0030] Based on the cumulative distribution function, a random number sequence between zero and one is generated to obtain a random number sequence.

[0031] Preferably, the steps of obtaining the resampled particle set are:

[0032] The random number sequence is traversed, and for each random number, all particles whose CDF values ​​are greater than the random number are identified by comparing the relationship with the cumulative distribution function, so as to form a new particle subset and an initial screening particle set;

[0033] For each particle in the initial screening particle set, an equal weight is reallocated to obtain a resampled particle set with a uniform weight, thereby forming a resampled particle set.

[0034] Preferably, the steps of obtaining the estimated value of the signal source position are:

[0035] Based on the resampled particle set, the product of the position information of each particle and the corresponding weight is calculated to obtain the weighted position information of each particle, thereby obtaining a weighted position information set;

[0036] The weighted position information set is accumulated, and the weights of all particles are accumulated at the same time, to obtain the sum of the weighted position information and the sum of the weights respectively.

[0037] Preferably, the step of acquiring the estimated value of the signal source position further comprises: dividing the sum of the weighted position information by the sum of the weights to calculate the estimated value of the signal source position to obtain the estimated position of the signal source.

[0038] Compared with the prior art, the advantages and positive effects of the present invention are:

[0039] The present invention achieves a wider detection range and higher positioning flexibility, generates randomly distributed position information in the area of ​​the initial position of the signal source, assigns an initial weight to each particle, and allows exploration in many possible discharge points. After combining multiple sensor data, the extracted signal strength and arrival time are used to update the weight of the particle, so that the weight of each particle more accurately reflects its positioning possibility. Resampling is performed through the cumulative distribution function of the particle weight, which optimizes the particle set and improves the calculation efficiency. The ratio of the sum of weighted position information to the sum of weights is accurately calculated, which improves the accuracy and efficiency of positioning. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 It is a schematic diagram of the steps of the present invention. DETAILED DESCRIPTION

[0041] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0042] See also Figure 1 The present invention provides a technical solution, a method for automatically locating a partial discharge signal, comprising the following steps:

[0043] According to the initial position area of ​​the partial discharge signal source, multiple groups of position information randomly distributed in the initial position area of ​​the partial discharge signal source are generated, the position information is assigned to each particle, the initial weight of each particle is calculated, and the corresponding weight is assigned to each particle to obtain an initial weighted particle set;

[0044] Based on the initial weighted particle set, signal data from multiple sensors are received, signal strength values ​​and arrival time values ​​are extracted, the predicted signal strength and arrival time of each particle position are compared with the actual values, the likelihood value of each particle is calculated, and the weight of the particle is updated by multiplying the original weight by the likelihood value to obtain an updated weighted particle set;

[0045] Based on the updated weighted particle set, the cumulative distribution function of the particle weight is calculated to generate a random number sequence between zero and one. For each random number, the particles whose cumulative distribution function is greater than the random number are captured, and the particles are selected to form a new particle set. The selected particles are given equal weights to obtain a resampled particle set.

[0046] Based on the resampled particle set, the position information of all particles is multiplied by the corresponding weights, the products are accumulated to obtain the sum of weighted position information, the weights of the particles are accumulated to obtain the sum of weights, and the sum of weighted position information is divided by the sum of weights to obtain the estimated value of the signal source position.

[0047] Specifically, randomly distributed position information in the area of ​​the initial position of the signal source is generated, and an initial weight is assigned to each particle, allowing exploration among many possible discharge points. After combining multiple sensor data, the extracted signal strength and arrival time are used to update the weight of the particle, so that the weight of each particle more accurately reflects its positioning possibility. Resampling is performed through the cumulative distribution function of the particle weight, which optimizes the particle set and improves the computational efficiency. The ratio of the sum of weighted position information to the sum of weights is accurately calculated to improve the accuracy and efficiency of positioning.

[0048] The steps to obtain the initial weighted particle set are:

[0049] According to the initial position area of ​​the partial discharge signal source, multiple sets of randomly distributed position information are generated, and each particle is assigned a position to obtain a position information set;

[0050] Based on the position information set, the initial weight is calculated for each particle, and the calculation formula is:

[0051] ;

[0052] in, is the weight of the ith particle, is the estimated distance from the particle to the signal source, is the expected distance offset, is the adjustment coefficient;

[0053] Based on the calculated weights, each particle is assigned a corresponding weight, and the initial weighted particle set is generated by combining the weights and position information.

[0054] Specifically, according to the initial location area of ​​the partial discharge signal source, based on the environmental characteristics in the area and the probability distribution of the signal source, multiple sets of randomly distributed location information are generated. This information represents the possible location of the partial discharge source. By arranging virtual particles in the simulation environment, each particle is assigned a random position. The obtained location information set provides a preliminary spatial data basis and provides the necessary input data for the subsequent weight allocation and signal source positioning.

[0055] The formula is useful in that by introducing the expected offset of the distance and adjustment coefficient , can more flexibly adjust the weight response curve, adapt to different signal attenuation models, and enhance positioning accuracy; Represents the estimated distance from the particle to the signal source, obtained based on the signal strength. The expected distance offset is pre-set according to the environmental noise and signal interference. To adjust the coefficient, its optimal value is determined through experimental data analysis to match the actual signal attenuation rate;

[0056] Calculation process: rice, rice, ,but:

[0057] ;

[0058] This result shows that when the particle is 40 meters away from the signal source, its weight is close to 1, which means that the probability of the particle at this distance is very high and its contribution to the positioning result is large;

[0059] According to the weight, each particle is assigned a corresponding weight, and the weight and position information are effectively combined. This process involves the normalization of the weight data and the weighted average calculation of the position data, ensuring that the weight of each particle accurately reflects its contribution to the positioning accuracy, forming an initial weighted particle set. This set contains all possible partial discharge source locations and their corresponding probability weights, which is fully prepared for the next step of signal source positioning, and obtains a preliminary, probability-weighted partial discharge signal source location estimate.

[0060] The specific steps for comparing the predicted signal strength and arrival time at each particle position with the actual values ​​and calculating the likelihood value of each particle are:

[0061] Receive signal data from multiple sensors, extract signal strength values ​​and arrival time values, and obtain initial signal data;

[0062] Based on the initial signal data, the predicted signal strength and arrival time of each particle position are compared with the actual values, and the likelihood value of each particle is calculated using the following formula:

[0063] ;

[0064] in, is the likelihood value of the ith particle, and are the predicted signal strength and arrival time, respectively. and is the observed signal strength and arrival time, is the standard deviation.

[0065] Specifically, in receiving signal data from multiple sensors, the data types collected by each sensor include signal strength and arrival time. The signal data is captured by the sensor in the actual environment. The signal strength is usually converted from the voltage change of the sensor, and the arrival time is recorded at the specific moment when the sensor first captures the signal. This process ensures that the data provided by each sensor is comparable and suitable for subsequent processing. After the data set is sorted, it includes comparative information on signal strength and arrival time collected by different sensors at the same time point.

[0066] The formula is beneficial in that it provides a way to quantify the accuracy of the prediction of the position of each particle by combining the errors in signal strength and arrival time; The acquisition step is to obtain the standard deviation of the statistical error of the data collected through multiple experiments in the early stage. and is the signal strength and arrival time predicted based on the particle model, and are observations extracted from actual sensor data.

[0067] Calculation process: predicted signal strength of the first particle at a certain moment , predicted arrival time Seconds; Observed signal strength , the observed arrival time Seconds; Standard Deviation . Substituting these values ​​into the formula:

[0068] ;

[0069] First calculate the standard deviation part in the denominator:

[0070] ;

[0071] Then calculate the numerator of the exponential part:

[0072] ;

[0073] Substituting these values ​​into the exponent part:

[0074] ;

[0075] Finally, put all the pieces together:

[0076] ;

[0077] The result shows that the predicted signal intensity and arrival time of the first particle match the actual observed values ​​by 0.447. The larger the value, the more consistent the prediction of the particle is with the observation, and vice versa.

[0078] The steps to update the weighted particle set are:

[0079] Based on the calculated likelihood value of each particle, the current weight of each particle is obtained to obtain the current weight set;

[0080] Based on the current weight set, multiply the current weight of each particle by the corresponding likelihood value, update the particle weight, and obtain the updated weight;

[0081] Based on the updated weights, the position information of each particle is recombined with the updated weights to generate an updated weighted particle set.

[0082] Specifically, the current weight of each particle is calculated, which involves comparing the likelihood value with preset performance parameters, which include safety thresholds, such as signal strength not less than 10 units or arrival time error not exceeding 2 milliseconds, to ensure that the weight of each particle is adjusted according to the accuracy of its prediction of the location of the PD signal source. This adjustment is based on the difference between the predicted value and the actual observed value. In this way, the weight of each particle is updated and determined to form the current weight set.

[0083] Based on the current weights obtained, the weights of each particle are updated by multiplication by their respective likelihood values. This calculation process involves numerical matching and adjustment of weights and likelihood values, especially ensuring that the weight values ​​are within a safe numerical range, such as not exceeding twice the original weights after the weight update. This is to avoid weight distortion due to extreme data points and ensure that each updated weight still reflects the effectiveness and importance of the particle in predicting the location of the PD signal source. In this way, an updated weight set is formed.

[0084] Based on the updated weights, a comprehensive reorganization of the particle position information and weights is performed. This step includes recalculating the weighted position of each particle using a weighted average method, in which the position coordinates of each particle are adjusted according to its updated weight to more accurately locate the signal source. The reorganization process ensures that the new particle set can respond to environmental changes more effectively based on the latest weight distribution. In this way, an updated weighted particle set is generated.

[0085] The steps to calculate the cumulative distribution function of particle weights and generate a sequence of random numbers between zero and one are:

[0086] Based on the updated weighted particle set, the updated weight of each particle is obtained to obtain an updated weight set;

[0087] Based on the updated weight set, calculate the cumulative distribution function of the particle weight , the calculation formula is:

[0088] ;

[0089] in, is the updated weight of the ith particle, is the position of the particle, is the target location, is the standard deviation, n is the total number of particles;

[0090] Based on the cumulative distribution function, a random number sequence between zero and one is generated to obtain a random number sequence.

[0091] Specifically, the formula is beneficial in that it simulates the relationship between particle weights and their positions by introducing a Gaussian function, which helps smooth the cumulative distribution function and improve the representativeness and fairness of random number generation; The parameter acquisition step is to obtain the signal strength and arrival time from the environmental sensor, calculate the distance between each particle and the signal source, and then determine it through the distance adjustment formula. The parameter represents the position of each particle, obtained from the sensor data, and σ represents the uncertainty of the position, which is the standard deviation estimated based on the environmental noise and signal attenuation;

[0092] Calculation process: First, for each particle , calculate its relative position to the target The square of the distance, that is , and then divide that value by , and taking the negative sign, we get Next, use this result as the exponential part of the exponential function to calculate the probability density value of the Gaussian distribution:

[0093] ;

[0094] Then, multiply the probability density value by the particle The updated weight , and get the weighted probability density value:

[0095] ;

[0096] Finally, the weighted probability density values ​​of all particles are added together to obtain the cumulative distribution function Values:

[0097] ;

[0098] This process ensures that the weights and positions of the particles are The contribution of is properly considered, so that the generated random number sequence can accurately reflect the distribution of particle weights.

[0099] Based on the cumulative distribution function, a random number sequence between zero and one is generated. This step is implemented through a random number generator, which generates random numbers according to the value distribution of the cumulative distribution function. These random numbers determine the selection process of subsequent particle resampling. The random number generator is designed to ensure the randomness and uniformity of the numbers to avoid deviations in the particle filtering process. The generation of this random sequence involves calling the computer's built-in random function and adjusting the probability density of the generated random numbers according to the distribution of CDF, so as to more accurately simulate the particle distribution and finally obtain a random number sequence.

[0100] The steps to obtain the resampled particle set are:

[0101] The random number sequence is traversed, and for each random number, all particles whose CDF value is greater than the random number are identified by comparing it with the cumulative distribution function, forming a new particle subset and an initial screening particle set;

[0102] For each particle in the initial screening particle set, an equal weight is redistributed to obtain a resampled particle set with uniform weight, thus forming a resampled particle set.

[0103] Specifically, by traversing the generated random number sequence one by one, the process of comparing each random number with the cumulative distribution function (CDF) involves accurately capturing particles whose CDF values ​​exceed the random value. This operation first requires calculating the CDF value of each particle, taking into account the position and weight of each particle, and obtaining a dynamically updated CDF chart. Based on this chart, particles that meet the conditions are screened out. These particles constitute a new particle subset, which is dynamically generated based on the relationship between random numbers and particle positions, ensuring that each selected particle can reflect the distribution characteristics of the entire particle group, thereby forming a representative initial screening particle set.

[0104] After obtaining the initial set of screened particles, the weights of each particle are equally redistributed. This step includes setting the weight of each particle to a uniform value. This process not only simplifies the complex calculation of weights, but also makes each particle have the same influence in subsequent processing, ensuring the fairness of the processing process.

[0105] The steps to obtain the estimated value of the signal source position are:

[0106] Based on the resampled particle set, the product of the position information of each particle and the corresponding weight is calculated to obtain the weighted position information of each particle, thereby obtaining a weighted position information set;

[0107] Accumulate the weighted position information set and the weights of all particles at the same time to obtain the sum of the weighted position information and the sum of the weights respectively;

[0108] The sum of the weighted position information is divided by the sum of the weights to calculate the estimated value of the signal source position and obtain the estimated position of the signal source.

[0109] Specifically, based on the resampled particle set, we first traverse all particles in the resampled particle set, identify and classify the position information of each particle, extract the corresponding weight value, and establish a matching table of particle positions and weights. Then, for each pair of positions and weights, we calculate the weighted position information of each particle one by one by performing a multiplication operation, and arrange all the calculation results according to the particle serial number to form a complete weighted position information set.

[0110] Each weighted value in the weighted position information set is accumulated item by item to form the sum of the weighted position information. At the same time, for all particles in the resampled particle set, their weight values ​​are accumulated separately to form the sum of the weights of all particles. To ensure the integrity of the accumulation process, all weighted values ​​and weight values ​​must be uniformly normalized before accumulation to ensure the consistency of the values, and finally form the sum of the position information and weight.

[0111] The sum of the weighted position information is ratioed with the sum of the weights to gradually calculate the estimated value of the signal source position. To ensure the accuracy of the estimated value, floating-point calculation method is used in the ratio operation, and the number of bits retained in the calculation result is adjusted to ensure that the accuracy of the final output signal source position estimate meets the requirements, and finally the estimated position of the signal source is obtained.

[0112] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. A method for automatically locating a partial discharge signal, characterized in that: The following steps are involved: According to the initial position area of ​​the partial discharge signal source, multiple groups of position information randomly distributed in the initial position area of ​​the partial discharge signal source are generated, the position information is assigned to each particle, the initial weight of each particle is calculated, and the corresponding weight is assigned to each particle to obtain an initial weighted particle set; Based on the initial weighted particle set, receiving signal data from multiple sensors, extracting signal strength values ​​and arrival time values, comparing the predicted signal strength and arrival time of each particle position with the actual values, calculating the likelihood value of each particle, and updating the particle weight by multiplying the original weight by the likelihood value to obtain an updated weighted particle set; Based on the updated weighted particle set, the cumulative distribution function of the particle weight is calculated to generate a random number sequence between zero and one. For each random number, the particles whose cumulative distribution function is greater than the random number are captured, the particles are selected to form a new particle set, and the selected particles are given equal weights to obtain a resampled particle set. Based on the resampled particle set, the position information of all particles is multiplied by the corresponding weights, the products are accumulated to obtain the sum of weighted position information, the weights of the particles are accumulated to obtain the sum of weights, and the sum of weighted position information is divided by the sum of weights to obtain an estimated value of the signal source position; The steps for obtaining the initial weighted particle set are: According to the initial position area of ​​the partial discharge signal source, multiple sets of randomly distributed position information are generated, and each particle is assigned a position to obtain a position information set; Based on the position information set, the initial weight is calculated for each particle, and the calculation formula is: ； in, is the weight of the ith particle, is the estimated distance from the particle to the signal source, is the expected distance offset, is the adjustment coefficient; Based on the calculated weights, each particle is assigned a corresponding weight, and an initial weighted particle set is generated by combining the weights and position information; The specific steps for comparing the predicted signal strength and arrival time at each particle position with the actual values ​​and calculating the likelihood value of each particle are: Receive signal data from multiple sensors, extract signal strength values ​​and arrival time values, and obtain initial signal data; Based on the initial signal data, the predicted signal strength and arrival time of each particle position are compared with the actual values, and the likelihood value of each particle is calculated using the following formula: ； in, is the likelihood value of the ith particle, and are the predicted signal strength and arrival time, respectively. and is the observed signal strength and arrival time, is the standard deviation; The steps to calculate the cumulative distribution function of particle weights and generate a sequence of random numbers between zero and one are: Based on the updated weighted particle set, obtaining an updated weight of each particle to obtain an updated weight set; Based on the updated weight set, calculate the cumulative distribution function of the particle weight , the calculation formula is: ； in, is the updated weight of the ith particle, is the position of the particle, is the target location, is the standard deviation, n is the total number of particles; Based on the cumulative distribution function, a random number sequence between zero and one is generated to obtain a random number sequence.

2. The method for automatically locating partial discharge signals according to claim 1, characterized in that: The acquisition steps of updating the weighted particle set are: Based on the calculated likelihood value of each particle, the current weight of each particle is obtained to obtain the current weight set; Based on the current weight set, multiply the current weight of each particle by the corresponding likelihood value, update the weight of the particle, and obtain an updated weight; Based on the updated weight, the position information of each particle is recombined with the updated weight to generate an updated weighted particle set.

3. The method for automatically locating partial discharge signals according to claim 1, characterized in that: The steps for obtaining the resampled particle set are: The random number sequence is traversed, and for each random number, all particles whose CDF values ​​are greater than the random number are identified by comparing the relationship with the cumulative distribution function, so as to form a new particle subset and an initial screening particle set; For each particle in the initial screening particle set, an equal weight is reallocated to obtain a resampled particle set with a uniform weight, thereby forming a resampled particle set.

4. The method for automatically locating partial discharge signals according to claim 1, characterized in that: The steps for obtaining the signal source position estimation value are as follows: Based on the resampled particle set, the product of the position information of each particle and the corresponding weight is calculated to obtain the weighted position information of each particle, thereby obtaining a weighted position information set; The weighted position information set is accumulated, and the weights of all particles are accumulated at the same time, to obtain the sum of the weighted position information and the sum of the weights respectively.

5. The method for automatically locating partial discharge signals according to claim 4, characterized in that: The step of acquiring the estimated value of the signal source position also includes: dividing the sum of the weighted position information by the sum of the weights to calculate the estimated value of the signal source position to obtain the estimated position of the signal source.

Images