Power distribution network reconstruction scheme optimization method and system based on visualization and power failure parameters

Abstract

The invention discloses a power distribution network reconstruction scheme optimization method and system based on visualization and power failure parameters, and belongs to the technical field of power distribution network construction. The method comprises the following steps: mapping a scalable vector graphic file to be verified to a general information model to generate a general information model to be verified; extracting a basic topological structure of the power distribution network, and dividing power equipment on a shortest path from a user node to a substation into necessary power equipment and non-necessary power equipment; constructing a LightGBM fault probability prediction model based on the influence of the necessary power equipment and the non-necessary power equipment on the power failure of the user, and calculating the power failure parameter of the user node based on the model; calculating power failure parameters of the transformer area system; and according to the power failure parameters of the user nodes and the power failure parameters of the transformer area system, solving to obtain an optimal power distribution network reconstruction scheme. The efficiency of the power distribution network transformation scheme is improved, the reliability and the real-time performance of power distribution network transformation are improved, manual intervention is reduced, and the comprehensive cost of power distribution network transformation is reduced.

Description

Technical Field

[0001] The present invention belongs to the technical field of distribution network construction, and more specifically, relates to an optimization method and system for a distribution network transformation plan based on visualization and power outage parameters. Background Art

[0002] The construction of a distribution network requires an analysis of the operation reliability of the distribution network. Based on the massive multi-source heterogeneous data provided by each operation scheduling monitoring system of the distribution network, considering the network structure of the distribution system and the parameters of each component, the real-time operation environment, the real-time operation state of the power grid, and factors such as the scheduling and maintenance plans, the operation index values of the system are given in real time.

[0003] An important criterion for the transformation of a distribution network is the ability of the power grid to continuously supply power, that is, the service ability of the power system to effectively provide electrical energy to users, which reflects the safety and stability of the power supply system. The transformation of the operating distribution network emphasizes more on real-time performance, fully considering the possible impact of the current device state and operation information in the actual environment on the stable operation in a future period. The prior art uses historical data for the transformation of the distribution network, ignoring the real-time data of the current device state and operation information in the actual environment, resulting in a mismatch between the historical device failure rate and the current real-time data, and further leading to unstable transformation of the distribution network.

[0004] In the prior art, it is relatively difficult to optimize the distribution network transformation plan. From the change in the topological structure to the quantification of the impact on users and the substation area system, it is a long and complex process. The time from drawing to verification of each plan is relatively long, and the workload of planners is relatively large. Summary of the Invention

[0005] To solve the deficiencies in the prior art, the present invention provides an optimization method and system for a distribution network transformation plan based on visualization and power outage parameters. By combining the distribution network grid topology information and the real-time device failure rate information, and then calculating the real-time power outage rate and power outage time indicators of load points and even the system level through LightGBM (Light Gradient Boosting Machine), and obtaining indicators such as the real-time equivalent annual power outage hours and average power outage duration of the distribution network system, thereby reflecting the feasibility of the current transformation plan of the system.

[0006] The present invention adopts the following technical solutions.

[0007] The first aspect of the present invention provides an optimization method for a distribution network transformation plan based on visualization and power outage parameters, including the following steps: Obtain the real-time scalable vector graphic file and real-time common information model of the current distribution network to be transformed in the substation area. Apply the transformation plan on the scalable vector graphic file to generate a scalable vector graphic file to be verified, and map the scalable vector graphic file to be verified to the common information model to generate a common information model to be verified; Extract the basic topology structure of the distribution network from the common information model to be verified, and divide the power equipment on the shortest path from the user node to the substation into essential power equipment and non-essential power equipment; Collect the historical operation data of the power equipment in the substation area within a set time range. Based on the impact of the essential power equipment and non-essential power equipment on user power outages, construct a LightGBM fault probability prediction model, and calculate the power outage parameters of the user node based on the model, including: expected power outage rate and expected power outage duration; Calculate the system power outage parameters of the substation area based on the power outage parameters of all user nodes in the substation area, including the expected average power outage frequency and expected power outage duration of the substation area; Based on the power outage parameters of the user nodes and the system power outage parameters of the substation area, construct a multi-objective optimization model with the goal of minimizing the power outage risk of users and the system, solve to obtain the optimal distribution network transformation plan, and realize the optimization of the distribution network transformation plan based on visualization and power outage parameters.

[0008] Preferably, the obtaining of the real-time scalable vector graphic file and real-time common information model of the current distribution network to be transformed in the substation area, applying the transformation plan on the scalable vector graphic file to generate a scalable vector graphic file to be verified, and mapping the scalable vector graphic file to be verified to the common information model to generate a common information model to be verified specifically includes: Obtain the scalable vector graphic file of the current distribution network to be transformed in the substation area, and convert it into an editable mode of editable elements. Preferably but not limited to, encode the scalable vector graphic file into an Extensible Markup Language format; In the editable mode, apply the transformation plan to the distribution network to be transformed, including: adding or deleting power equipment and / or power lines to form a topology structure of the distribution network to be verified; after the compilation is completed, generate a scalable vector graphic file to be verified; Obtain the common information model of the current distribution network to be transformed, and map the topology structure of the distribution network to be verified in the scalable vector graphic file to be verified to the common information model in a linked list structure to generate a common information model to be verified.

[0009] Preferably, the linked list structure includes the type of power equipment added or deleted in the transformation plan, the input end identification code, the output end identification code, the power equipment model, and the power line length.

[0010] Preferably, in the general information model to be verified, extract the basic topology of the distribution network, and divide the power equipment on the shortest path from the user node to the substation into essential power equipment and non-essential power equipment, which specifically includes: Extract the basic topology of the distribution network from the general information model to be verified to obtain the basic topology diagram of the distribution network to be verified; Based on the basic topology diagram of the distribution network to be verified, start traversing and searching for the shortest path from each user to the substation according to the distance between the user node and the substation. According to all the shortest paths, divide all the power equipment in the basic topology diagram of the distribution network into essential power equipment and non-essential power equipment. The power equipment that appears in any shortest path is essential power equipment, and the power equipment that does not appear in any shortest path is non-essential power equipment.

[0011] Preferably, collect the historical operation data of the power equipment in the area within a set time range, build a LightGBM fault probability prediction model based on the impact of essential power equipment and non-essential power equipment on user power outages, and calculate the power outage parameters of the user node based on the model, which specifically includes: Collect the historical operation data of the power equipment in the area within a set time range and perform preprocessing; Train the LightGBM fault probability prediction model based on the preprocessed dataset to obtain the trained LightGBM model; Input the real-time device status and environmental factors into the trained LightGBM model, dynamically update the posterior probability through the Markov chain Monte Carlo method, and obtain the device fault probability matrix according to the essential power equipment and non-essential power equipment; Conduct Monte Carlo simulation based on the device fault probability matrix to calculate the expected power outage rate of the user node and the expected power outage duration of the user node.

[0012] Preferably, the training of the LightGBM fault probability prediction model based on the preprocessed dataset to obtain the trained LightGBM model specifically includes: Based on the logarithmic loss function, with the goal of minimizing the difference between the predicted fault probability and the actual label, construct the objective function of the LightGBM fault probability prediction model and solve the loss function; Optimize the splitting nodes and leaf weights of each decision tree for the device sample status feature values through the gradient descent method to generate decision trees; Calculate the weight of each tree through the combination of the decision tree, the loss function, and backpropagation; When the loss function of the validation set does not decrease for K consecutive rounds, terminate the training to obtain the trained LightGBM model, and save the optimal tree set and the optimal weight.

[0013] Preferably, by combining decision trees with loss functions and backpropagation, calculate the weights of each tree, specifically including: For each sample i, calculate the prediction error; Combined with the loss value, multiply the prediction error by the output value of the t-th tree, accumulate the results of all samples, and obtain the gradient of the loss function with respect to the weight of the t-th tree; Calculate the weights of each tree according to the gradient.

[0014] Preferably, input the real-time device status and environmental factors into the trained LightGBM model, dynamically update the posterior probability through the Markov chain Monte Carlo method, and obtain the device failure probability matrix according to the essential power equipment and non-essential power equipment, specifically including: Set the Bayesian prior for the optimal weight of each tree to obtain the initialized posterior distribution; Update the weight parameters through the Metropolis-Hastings algorithm based on the initialized posterior distribution and the real-time device status and environmental factors; Calculate the acceptance probability based on the candidate parameters, compare the joint probabilities of the candidate parameters and the current parameters, determine whether to accept the new parameters, and set the weights obtained from the judgment result as the updated weights; According to the updated weights, solve the predicted failure probability and set it as the basic failure probability. Combine the essential power equipment and non-essential power equipment, with the device ID as the row and the device status as the column, and fill in the failure probability of the device at the corresponding position to generate the failure probability matrix.

[0015] Preferably, perform Monte Carlo simulation based on the device failure probability matrix to calculate the expected power outage rate of the user node and the expected power outage duration of the user node, specifically including: Conduct independent sampling for each device j, generate a random number according to the failure probability value corresponding to the device status, and determine it as a failure if it is less than the failure probability; Statistically calculate the failure probabilities of all devices, and perform weighted summation according to the reciprocal of the number of path devices to obtain the expected power outage rate of the user node; Statistically calculate the repair times of all failed devices, and perform weighted summation according to the reciprocal of the number of path devices to obtain the expected power outage duration of the user node.

[0016] The second aspect of the present invention provides a distribution network transformation plan optimization system based on visualization and power outage parameters, which runs the distribution network transformation plan optimization method based on visualization and power outage parameters described in the first aspect, including: A data acquisition module for obtaining the scalable vector graphic file, common information model of the currently to-be-transformed distribution network in the substation area, and the historical operation data of the power equipment in the substation area within a set time range; A visual data conversion module is used to apply a transformation scheme on the basis of the scalable vector graphics file to generate a scalable vector graphics file to be verified, and map the scalable vector graphics file to be verified to a general information model to generate a general information model to be verified; The equipment classification module is used to extract the basic topology of the distribution network in the general information model to be verified, and classify the power equipment on the shortest path from the user node to the substation into necessary power equipment and non-necessary power equipment; The user node power outage parameter solution module is used to collect the historical operation data of the power equipment in the substation within the set time range, build the LightGBM fault probability prediction model based on the impact of the necessary power equipment and non-necessary power equipment on the user's power outage, and calculate the power outage parameters of the user node based on the model, including: the expected power outage rate and the expected power outage duration; The power outage parameter solving module of the substation system is used to calculate the power outage parameters of the substation system based on the power outage parameters of all user nodes in the substation, including the expected average power outage frequency and expected power outage duration of the substation; The output module is used to construct a multi-objective optimization model based on the power outage parameters of the user node and the power outage parameters of the substation system, with the goal of minimizing the risk of power outages for users and the system, to solve the optimal distribution network transformation plan, and to achieve the optimization of the distribution network transformation plan based on visualization and power outage parameters.

[0017] Compared with the prior art, the beneficial effects of the present invention include at least: Generate transformation plans through visual editing of vector graphics files, automatically map them to the general information model, save programming steps, improve efficiency, build a multi-objective optimization model, automatically quantify the impact of the plan on users and system power outage parameters, replace manual trial and error, simplify the transformation plan design process, and improve the efficiency of distribution network transformation plans; Since the failure of the necessary equipment on the main feeder will cause power outages for all users on the path, the failure of non-necessary equipment will only affect a local area. Based on the shortest path, the necessary equipment and non-necessary equipment are divided. The key risks are amplified by the necessary equipment weight correction factor introduced by the necessary equipment and the success probability of the isolation device introduced by the non-necessary equipment to quantify the effect of the protection device, improve the accuracy of fault prediction, and reduce the calculation error of user-level power outage parameters; Based on the LightGBM model and real-time data, the fault probability is dynamically predicted, combined with the Markov chain Monte Carlo to update the posterior probability, and the latest data is integrated in real time to improve the reliability of the model's posterior probability and the accuracy of fault prediction. The Monte Carlo simulation is used to calculate the power outage parameters in real time, which can dynamically reflect the impact of the current equipment status, adapt to environmental changes, and improve the stability, reliability and real-time performance of the distribution network transformation. The multi-objective model automatically screens the optimal solution, balances the risks of user-level and system-level power outages, avoids excessive redundant design, reduces manual intervention, and lowers the overall cost of distribution network transformation. Description of the Drawings

[0018] Figure 1 is a flowchart of an optimization method for a distribution network transformation plan based on visualization and real-time power outage parameters provided according to an embodiment of the present invention; Figure 2 is a schematic diagram of extracting the basic topological structure of a distribution network in a general information model to be verified provided according to an embodiment of the present invention. Detailed Embodiment

[0019] To make the objectives, technical solutions, and advantages of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. The described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the spirit of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0020] As Figure 1 shown, Embodiment 1 of the present invention provides an optimization method for a distribution network transformation plan based on visualization and real-time power outage parameters, including the following steps: Step 1: Obtain a scalable vector graphics file of the distribution network to be transformed in the current substation area, apply a transformation plan based on the scalable vector graphics file to generate a scalable vector graphics file to be verified, and map the scalable vector graphics file to be verified to a general information model to generate a general information model to be verified.

[0021] In a preferred but non-limiting embodiment of the present invention, Step 1 specifically includes: Step 1.1: Obtain a scalable vector graphics file of the distribution network to be transformed in the current substation area and convert it into an editable mode of editable elements. Preferably but not limited to, encode the scalable vector graphics file into an Extensible Markup Language format.

[0022] Step 1.2: In the editable mode, apply the transformation plan to the distribution network to be transformed, including adding or deleting power equipment and / or power lines to form a topological structure of the distribution network to be verified; after the compilation is completed, generate a scalable vector graphics file to be verified.

[0023] Step 1.3: Obtain the general information model of the distribution network to be transformed in the current substation area, map the topological structure of the distribution network to be verified in the scalable vector graphics file to be verified to the general information model in a linked list structure, and generate a general information model to be verified.

[0024] Preferably, the linked list structure includes parameters such as the type of power equipment added or deleted in the transformation plan, the input end identification code, the output end identification code, the power equipment model, and the power line length.

[0025] It should be noted that adding or deleting power equipment and / or power lines can be carried out in a scalable vector graphics file rather than on the basis of a common information model, and can be carried out in a visual manner without having to be entered in the form of programming language code, improving the work efficiency of planners and designers.

[0026] Step 2: In the to-be-verified common information model, extract the basic topology structure of the distribution network, and divide the power equipment on the shortest path from the user node to the substation into essential power equipment and non-essential power equipment. As Figure 2 shown.

[0027] In a preferred but non-limiting embodiment of the present invention, Step 2 specifically includes: Step 2.1: In the to-be-verified common information model, extract the basic topology structure of the distribution network, including: busbars, circuit breakers, disconnecting switches, distribution transformers, overhead lines, and cables, to obtain the to-be-verified basic topology structure diagram of the distribution network.

[0028] Specifically, extracting the basic topology structure of the distribution network includes: Merging elements: Merge the same-type and adjacent cables and overhead lines between the set nodes, including: between the starting point and the branch node, between branch nodes, and between the branch node and the end point.

[0029] Ignoring elements: Ignore power equipment such as oscillographs, current transformers, and lightning arresters.

[0030] It can be understood that in the power system, the scale of the complete common information model is very large, including a series of power system models such as a core package, a topology package, and a protection package, which are specified in detail in DL / T 890.301-2016. During the verification of the distribution network transformation, it is not necessary to load all models. The present invention only extracts the key information, and this key information is newly mapped from the scalable vector graphics file to the common information model, and the verification is achieved with only a very small workload.

[0031] Step 2.2: On the basis of the to-be-verified basic topology structure diagram of the distribution network, start traversing and searching for the shortest path from each user to the substation according to the distance between the user node, that is, the load node, and the substation. According to all the shortest paths, divide all the power equipment in the basic topology structure diagram of the distribution network into essential power equipment and non-essential power equipment. The power equipment that appears in any shortest path is essential power equipment, and the power equipment that does not appear in any shortest path is non-essential power equipment.

[0032] Without considering the standby power supply, the outage of a necessary power equipment will inevitably lead to the power outage of users on the shortest path where it is located; the outage of a non-necessary power equipment does not necessarily cause the power outage of its downstream users. Further, if the non-necessary power equipment is located on a branch line and a fault isolation device is installed upstream of it, the outage of this non-necessary power equipment will not cause the power outage of other branch lines; if the non-necessary power equipment is located on the main feeder, the outage of the non-necessary power equipment downstream of the disconnecting switch or sectionalizing circuit breaker will cause the power outage time of its upstream users to be only the operation time of the disconnecting switch or sectionalizing circuit breaker.

[0033] Step 3: Collect the historical operation data of power equipment in the substation area within a set time range. Based on the impact of necessary power equipment and non-necessary power equipment on user power outages, construct a LightGBM fault probability prediction model, and calculate the power outage parameters of user nodes based on the model, including: expected power outage rate and expected power outage duration.

[0034] In a preferred but non-limiting embodiment of the present invention, Step 3 specifically includes: Step 3.1: Collect the historical operation data of power equipment in the substation area within a set time range and perform preprocessing. Among them, the historical operation data of power equipment includes the equipment historical state sequence, environmental data, and fault records.

[0035] Further preferably, Step 3.1 includes: Step 3.1.1: Collect the historical operation data of power equipment in the substation area within a set time range.

[0036] More preferably, Step 3.1.1 includes: Collect the equipment historical state sequence, including the states of power equipment operating normally, power equipment that needs attention in operation, power equipment operating abnormally, and power equipment with major abnormalities; Collect environmental data, such as environmental parameters of meteorological conditions, temperature, humidity, etc.; Collect the fault records of power equipment. The time range is preferably but not limited to the most recent year.

[0037] Step 3.1.2: Perform preprocessing on the historical operation data of power equipment.

[0038] More preferably, Step 3.1.2 includes: Convert the equipment historical state sequence into numerical labels, and add binary features according to the fault records of power equipment S 关键性 , identify whether the equipment is a necessary equipment, 1 for yes, otherwise 0, to obtain the state characteristic values of equipment samples S = [S 1 ,S 2 ,S 3 , S 4,S 关键性 ] , where S 1 represents the normal operating state of power equipment, S 2 represents the state of power equipment that requires attention to the operating state, S 3 represents the abnormal operating state of power equipment, S 4 represents the state of power equipment with major abnormalities; Normalize the meteorological data according to the disaster level to obtain the environmental factor ; Combine the state characteristic values of the combined equipment samples and the environmental factor to obtain the preprocessed data set, including the feature matrix .

[0039] Step 3.2: Based on the preprocessed data set, train the LightGBM fault probability prediction model to obtain the trained LightGBM model, including the tree ensemble {ht} and the weights {βt}.

[0040] Further preferably, step 3.2 includes: Step 3.2.1, based on the logarithmic loss function, with the goal of minimizing the difference between the predicted fault probability and the actual label, construct the objective function of the LightGBM fault probability prediction model, and solve the loss value, which is expressed by the following formula:

[0041] In the formula, is the loss value, is the total number of training samples; is the true label of the i-th sample, 0 means no fault, and 1 means fault; is the fault probability of the i-th sample predicted by the model.

[0042] Step 3.2.2, optimize the split nodes and leaf weights of each decision tree for the state characteristic values of the equipment samples by the gradient descent method to generate decision trees.

[0043] Step 3.2.3, calculate the weight of each tree through the decision tree combined with the loss function and backpropagation, specifically including: For each sample i, calculate the prediction error ; Combine the loss value, multiply the prediction error by the output value of the t-th tree, accumulate the results of all samples to obtain the gradient , and calculate the weight of each tree according to the gradient, which is expressed by the following formula:

[0044] In the formula, represents the loss function for the weight of the k-th tree; represents the true label of the i-th sample, 0 means no fault, and 1 means fault; represents the output value of the k-th decision tree for sample i; represents the device state of the i-th sample; represents the predicted fault probability of the i-th sample, expressed by the following formula:

[0045] In the formula, represents the Sigmoid function; represents the number of decision trees; represents the weight of the t-th tree in the k-th iteration;

[0046] Step 3.2.4, when the validation set loss function has not decreased for K consecutive rounds, terminate the training to obtain the trained LightGBM model, and save the optimal tree set and the optimal weights.

[0047] Step 4, input the real-time device state and environmental factors into the trained LightGBM model, dynamically update the posterior probability through the Markov chain Monte Carlo method, and obtain the device fault probability matrix according to the essential and non-essential power equipment.

[0048] In a preferred but non-limiting embodiment of the present invention, Step 4 includes: Step 4.1, perform Bayesian prior setting for the optimal weights of each tree to obtain the initialized posterior distribution, expressed by the following formula:

[0049] In the formula, is the initialized posterior distribution, is the normal distribution, is the mean value, and the initially trained weights are set to the mean value, is the variance.

[0050] Step 4.2: Update the weight parameters according to the initialized posterior distribution, real-time device status, and environmental factors through the Metropolis-Hastings algorithm, specifically including: Generate a proposal distribution centered on the current weights, and generate candidate parameters according to the variance, which is expressed by the following formula:

[0051] In the formula, is the candidate parameter; is the variance of the proposal distribution.

[0052] Step 4.3: Calculate the acceptance probability according to the candidate parameters, compare the joint probabilities of the candidate parameters and the current parameters, determine whether to accept the new parameters, and set the weights obtained from the judgment result as the updated weights, which is expressed by the following formula:

[0053] In the formula, is the acceptance probability, is the likelihood function, indicating the probability of observing new data under the current weights .

[0054] If , accept , otherwise retain , where represents the uniform distribution.

[0055] Step 4.4: According to the updated weights, solve the predicted fault probability according to Formula 4 and set it as the basic fault probability. Combine the essential power equipment and non-essential power equipment, use the equipment ID as the row and the equipment status as the column, and fill in the fault probability of the equipment at the corresponding position to generate a fault probability matrix, which is expressed by the following formula:

[0056] In the formula, represents the fault probability of the essential equipment; represents the fault probability of the non-essential equipment; represents the basic fault probability; Represents the weight correction factor of the mandatory device; Represents the success probability of the isolation device.

[0057] Step 5: Conduct Monte Carlo simulation based on the failure probability matrix, calculate the power outage parameters of user nodes, and calculate the power outage parameters of the substation area system based on the power outage parameters of all user nodes in the substation area, including the expected average power outage frequency and the expected power outage duration of the substation area.

[0058] In a preferred but non-limiting embodiment of the present invention, Step 5 includes: Step 5.1, conduct independent sampling for each device j, generate a random number according to the failure probability value corresponding to the device state, if it is less than the failure probability, it is determined as a failure, and solve the expected power outage rate and the expected power outage duration of the user node, which are expressed by the following formula:

[0059] In the formula, is the expected power outage rate of the user node ; is the user node the set of devices on the shortest path to the substation; is the failure probability of device j; is the user node the expected power outage duration; is the device historical repair time.

[0060] Step 5.2, calculate the expected average power outage frequency and the expected power outage duration of the substation area based on the users, which are expressed by the following formula:

[0061] In the formula: is the expected average power outage frequency of the substation area; is the expected power outage duration of the substation area; is the total number of user nodes in the substation area, .

[0062] Step 6: Based on the power outage parameters of user nodes and the power outage parameters of the substation area system, a multi-objective optimization model is constructed with the goal of minimizing the power outage risk of users and the system. The optimal distribution network renovation plan is obtained through solution, realizing the optimization of the distribution network renovation plan based on visualization and power outage parameters, which is expressed by the following formula:

[0063] In the formula: is the objective function of the renovation plan, aiming to comprehensively reduce the power outage parameters of users and the substation area; is the expected power outage rate of user node before renovation; is the expected power outage duration of user node before renovation; are the constraint conditions; are the weights of the first and second power outage rates, are the weights of the first and second power outage times.

[0064] Embodiment 2 of the present invention provides a system for optimizing a distribution network renovation plan based on visualization and power outage parameters, which runs the method for optimizing a distribution network renovation plan based on visualization and power outage parameters described in Embodiment 1, including: A data acquisition module, which is used to obtain the scalable vector graphics file, the common information model of the currently to-be-renovated distribution network in the substation area, and the historical operation data of the power equipment in the substation area within a set time range; A visualization data conversion module, which is used to apply the renovation plan on the basis of the scalable vector graphics file to generate a to-be-verified scalable vector graphics file, and map the to-be-verified scalable vector graphics file to the common information model to generate a to-be-verified common information model; An equipment division module, which is used to extract the basic topological structure of the distribution network in the to-be-verified common information model, and divide the power equipment on the shortest path from the user node to the substation into necessary power equipment and non-necessary power equipment; A user node power outage parameter solving module, which is used to collect the historical operation data of the power equipment in the substation area within a set time range, construct a LightGBM fault probability prediction model based on the impact of necessary power equipment and non-necessary power equipment on user power outages, and calculate the power outage parameters of user nodes based on the model, including: the expected power outage rate and the expected power outage duration; A substation area system power outage parameter solving module, which is used to calculate the power outage parameters of the substation area system based on the power outage parameters of all user nodes in the substation area, including the expected average power outage frequency and the expected power outage duration of the substation area; An output module, which is used to construct a multi-objective optimization model with the goal of minimizing the power outage risks of users and the substation area system according to the power outage parameters of user nodes and the substation area system, solve to obtain an optimal distribution network transformation plan, and realize the optimization of the distribution network transformation plan based on visualization and power outage parameters.

[0065] Compared with the prior art, the beneficial effects of the present invention at least include: By generating a transformation plan through visual editing of vector graphics files and automatically mapping it to a common information model, the programming steps are omitted, the efficiency is improved, a multi-objective optimization model is constructed, the impact of the plan on user and system power outage parameters is automatically quantified, manual trial and error is replaced, the design process of the transformation plan is simplified, and the efficiency of the distribution network transformation plan is improved; Since a power outage of all users on the path will be caused by a failure of a necessary device on the main feeder, while a failure of a non-necessary device only affects a local area, the necessary and non-necessary devices are divided based on the shortest path. The key risk is amplified by the necessary device weight correction factor introduced by the necessary device, and the success probability of the isolation device introduced by the non-necessary device is quantified to evaluate the protection device effect, improving the accuracy of fault prediction and reducing the calculation error of user-level power outage parameters; Based on the LightGBM model and real-time data, the fault probability is dynamically predicted, combined with the Markov chain Monte Carlo to update the posterior probability, and the latest data is fused in real time to improve the reliability of the model's posterior probability and the accuracy of fault prediction. By Monte Carlo simulation, the power outage parameters are calculated in real time, which can dynamically reflect the impact of the current device state, adapt to environmental changes, and improve the reliability and real-time performance of the distribution network transformation; The multi-objective model automatically screens the optimal plan, balances the power outage risks at the user level and the system level, avoids excessive redundant design, reduces manual intervention, and reduces the comprehensive cost of the distribution network transformation.

[0066] The present disclosure may be a system, a method, and / or a computer program product. The computer program product may include a computer-readable storage medium having computer-readable program instructions thereon for causing a processor to implement various aspects of the present disclosure.

[0067] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the above embodiments, those of ordinary skill in the art should understand that: the specific implementation manners of the present invention can still be modified or equivalently replaced, and any modification or equivalent replacement without departing from the spirit and scope of the present invention shall be covered by the protection scope of the claims of the present invention.

Claims

1. An optimization method for the transformation plan of a distribution network based on visualization and power outage parameters, characterized in that, The method includes the following steps: Obtain the real-time scalable vector graphic file and the real-time common information model of the current distribution network to be transformed in the substation area. Apply the transformation plan to the scalable vector graphic file to generate a scalable vector graphic file to be verified, and map the scalable vector graphic file to be verified to the common information model to generate a common information model to be verified; Extract the basic topology structure of the distribution network from the common information model to be verified, and divide the power equipment on the shortest path from the user node to the substation into essential power equipment and non-essential power equipment; Collect the historical operation data of the power equipment in the substation area within a set time range, and construct a LightGBM fault probability prediction model based on the impact of the essential power equipment and non-essential power equipment on user power outages; Input the real-time device status and environmental factors into the trained LightGBM model to obtain a device fault probability matrix; Calculate the power outage parameters of the user nodes based on the device fault probability matrix, and calculate the power outage parameters of the substation area system based on the power outage parameters of all user nodes in the substation area. Among them, the power outage parameters of the user nodes include the expected power outage rate and the expected power outage duration, and the power outage parameters of the substation area system include the expected average power outage frequency and the expected power outage duration of the substation area; Based on the power outage parameters of the user nodes and the power outage parameters of the substation area system, construct a multi-objective optimization model with the goal of minimizing the power outage risk of users and the system, solve to obtain the optimal distribution network transformation plan, and realize the optimization of the distribution network transformation plan based on visualization and power outage parameters.

2. The optimization method for the distribution network transformation plan based on visualization and power outage parameters according to claim 1, wherein: The step of obtaining the real-time scalable vector graphic file and the real-time common information model of the current distribution network to be transformed in the substation area, applying the transformation plan to the scalable vector graphic file to generate a scalable vector graphic file to be verified, and mapping the scalable vector graphic file to be verified to the common information model to generate a common information model to be verified specifically includes: Obtain the scalable vector graphic file of the current distribution network to be transformed in the substation area, and convert it into an editable mode of editable elements; In the editable mode, apply the transformation plan to the distribution network to be transformed, including adding and deleting power equipment and / or power lines to form a topology structure of the distribution network to be verified; after the compilation is completed, generate a scalable vector graphic file to be verified; Obtain the common information model of the current distribution network to be transformed, and map the topology structure of the distribution network to be verified in the scalable vector graphic file to be verified to the common information model in a linked list structure to generate a common information model to be verified.

3. The optimization method for the distribution network transformation plan based on visualization and power outage parameters according to claim 2, wherein: The step of extracting the basic topology structure of the distribution network from the common information model to be verified and dividing the power equipment on the shortest path from the user node to the substation into essential power equipment and non-essential power equipment specifically includes: Extract the basic topology structure of the distribution network from the common information model to be verified to obtain a basic topology structure diagram of the distribution network to be verified; Based on the basic topological structure diagram of the distribution network to be verified, starting from the distance between user nodes and substations, traverse and search for the shortest path from each user to the substation. According to all the shortest paths, all power equipment in the basic topological structure diagram of the distribution network is divided into necessary power equipment and non-necessary power equipment. The power equipment that appears in any shortest path is set as necessary power equipment, and the power equipment that does not appear in any shortest path is set as non-necessary power equipment.

4. The optimization method for the distribution network transformation plan based on visualization and power outage parameters according to claim 1, characterized in that: Collect the historical operation data of the power equipment in the area within the set time range, and construct a LightGBM fault probability prediction model based on the impact of necessary power equipment and non-necessary power equipment on user power outages, specifically including: Collect the historical operation data of the power equipment in the area within the set time range and perform preprocessing; Train the LightGBM fault probability prediction model based on the preprocessed data set to obtain the trained LightGBM model.

5. The optimization method for the distribution network transformation plan based on visualization and power outage parameters according to claim 4, characterized in that: The training of the LightGBM fault probability prediction model based on the preprocessed data set to obtain the trained LightGBM model specifically includes: Based on the logarithmic loss function, with the goal of minimizing the difference between the predicted fault probability and the actual label, construct the objective function of the LightGBM fault probability prediction model and solve the loss function; Optimize the splitting nodes and leaf weights of each decision tree through the gradient descent method with the device sample state eigenvalue to generate decision trees; Calculate the weight of each tree through the combination of decision trees, loss function, and backpropagation; When the loss function of the validation set has not decreased for K consecutive rounds, terminate the training to obtain the trained LightGBM model, and save the optimal tree set and optimal weights.

6. The optimization method for the distribution network transformation plan based on visualization and power outage parameters according to claim 5, characterized in that: Calculating the weight of each tree through the combination of decision trees, loss function, and backpropagation specifically includes: For each sample i, calculate the prediction error; Combined with the loss value, multiply the prediction error by the output value of the t-th tree, accumulate the results of all samples, and obtain the gradient of the loss function with respect to the weight of the t-th tree; Calculate the weight of each tree according to the gradient.

7. The optimization method for the distribution network transformation plan based on visualization and power outage parameters according to claim 6, characterized in that: Input the real-time device status and environmental factors into the trained LightGBM model to obtain the device fault probability matrix, specifically including: Input the real-time device status and environmental factors into the trained LightGBM model, dynamically update the posterior probability through the Markov chain Monte Carlo method, and obtain the device fault probability matrix according to the necessary power equipment and non-necessary power equipment.

8. The optimization method for the distribution network transformation plan based on visualization and power outage parameters according to claim 7, characterized in that: Input the real-time device status and environmental factors into the trained LightGBM model, dynamically update the posterior probability through the Markov chain Monte Carlo method, and obtain the device failure probability matrix according to the essential power equipment and non-essential power equipment, specifically including: Set the Bayesian prior for the optimal weight of each tree to obtain the initialized posterior distribution; Update the weight parameters according to the initialized posterior distribution, real-time device status and environmental factors through the Metropolis-Hastings algorithm; Calculate the acceptance probability based on the candidate parameters, compare the joint probability of the candidate parameters and the current parameters, determine whether to accept the new parameters, and set the weight obtained from the judgment result as the updated weight; Solve the predicted failure probability based on the updated weight and set it as the basic failure probability. Combine the essential power equipment and non-essential power equipment, use the device ID as the row and the device status as the column, and fill in the failure probability of the device at the corresponding position to generate the failure probability matrix.

9. The optimization method for the distribution network transformation plan based on visualization and power outage parameters according to claim 7, characterized in that: Calculating the power outage parameters of the user node based on the device failure probability matrix specifically includes: Independently sample each device j, generate a random number according to the failure probability value corresponding to the device status, and determine it as a failure if it is less than the failure probability; Statistically calculate the failure probabilities of all devices, and perform weighted summation according to the reciprocal of the number of path devices to obtain the expected power outage rate of the user node; Statistically calculate the repair times of all failed devices, and perform weighted summation according to the reciprocal of the number of path devices to obtain the expected power outage duration of the user node.

10. An optimization system for the distribution network transformation plan based on visualization and power outage parameters, which runs the optimization method for the distribution network transformation plan based on visualization and power outage parameters according to any one of claims 1-9, characterized in that: A data acquisition module for obtaining the scalable vector graphic file, common information model of the current distribution network to be transformed in the substation area, and the historical operation data of the power equipment in the substation area within a set time range; A visualization data conversion module for applying the transformation plan on the basis of the scalable vector graphic file to generate a scalable vector graphic file to be verified, and mapping the scalable vector graphic file to be verified to the common information model to generate a common information model to be verified; A device division module for extracting the basic topological structure of the distribution network in the common information model to be verified, and dividing the power equipment on the shortest path from the user node to the substation into essential power equipment and non-essential power equipment; A user node power outage parameter solving module for collecting the historical operation data of the power equipment in the substation area within a set time range, constructing a LightGBM failure probability prediction model based on the impact of essential power equipment and non-essential power equipment on user power outages, and calculating the power outage parameters of the user node based on the model, including: expected power outage rate and expected power outage duration; A substation area system power outage parameter solving module for calculating the substation area system power outage parameters based on the power outage parameters of all user nodes in the substation area, including the expected average power outage frequency and expected power outage duration of the substation area; An output module, which is used to construct a multi-objective optimization model with the goal of minimizing the power outage risks of users and the district system according to the power outage parameters of user nodes and the district system power outage parameters, solve to obtain the optimal distribution network transformation plan, and realize the optimization of the distribution network transformation plan based on visualization and power outage parameters.

Images