CN-122022453-A - KMP method-based power grid asset wall risk prediction algorithm and system

CN122022453ACN 122022453 ACN122022453 ACN 122022453ACN-122022453-A

Abstract

The invention aims to provide a grid asset wall risk prediction algorithm and system based on a KMP method. According to the invention, through measuring and calculating the wall-in years of various devices of the power grid, the wall-in scale of the asset is predicted, and the risk of the asset wall is predicted. Firstly, calculating a reference rejection probability curve by using a K-M method, then fitting external influence factors by using Prophet, correcting the reference rejection probability curve, sequentially calculating the in-wall years of various assets of the power grid, finally calculating the in-wall asset scale and value of the target year, predicting the future asset wall scale of the power grid, and judging the risk level of the asset wall. According to the method, on the basis of considering the average service life of various equipment of the power grid, the influence of external influence factors on the asset rejection probability is also considered, the service life distribution model is respectively built for various equipment, the in-wall asset value of the equipment is predicted, the dynamic and fine prediction of the asset wall risk is realized, and the risk prediction precision and timeliness of various and long-service-life assets of the power grid are obviously improved.

Inventors

LI XUYANG
NIU XIN
LI PENG
GUO ZHENGWEI
LI YONGJIE
LI LINGYUN
XU YANG
WANG JING
ZHOU TIEJUN

Assignees

国网河南省电力公司经济技术研究院

Dates

Publication Date: 20260512
Application Date: 20251231

Claims (8)

1. The utility model provides a grid asset wall risk prediction algorithm based on KMP method which is characterized by comprising the following steps: Step 1, collecting a historical retired scrapped asset list and a current in-service asset list of a power grid, and calculating a reference scrapped amount according to the age of the retired scrapped asset of the power grid to obtain a reference scrapped probability time sequence; Step 2, collecting external factor historical data, wherein the external factor historical data comprises, but is not limited to, environmental influence, running state and operation and maintenance strategy, predicting external influence factors which fluctuate with time and influence the reference rejection probability, and obtaining an environmental pressure index time sequence; step 3, correcting the reference scrapping probability time sequence in the step 1 by using the environmental pressure index time sequence in the step 2 to obtain the final dynamically adjusted asset scrapping risk; step 4, determining the in-wall years of various assets through preset asset rejection thresholds; Step 5, traversing all in-service assets in the power grid, calculating the value of the in-wall assets in the designated year, and measuring and calculating the risk scale of the asset wall; and 6, predicting the investment capacity of the power grid in the designated year, calculating the ratio relation between the risk scale of the asset wall and the investment capacity of the power grid, and determining the risk level of the asset wall.
2. The grid asset wall risk prediction algorithm based on the KMP method according to claim 1, wherein the calculation process of the reference rejection probability time sequence in the step 1 is as follows: step 1.1, collecting a historical retired scrapped asset list and a current in-service asset list of a power grid: The historical retired scrapped asset list comprises, but is not limited to, asset codes, asset names, asset classifications, asset original values, technical scales, commissioning dates, retired scrapping dates and retired scrapping reason fields, and the current in-service asset list comprises, but is not limited to, asset codes, asset names, asset classifications, asset original values, technical scales, commissioning dates and running state fields, wherein the technical scales comprise, but are not limited to, the length of line type assets and the capacity of power transformation type assets. Step 1.2, calculating the ages of all the assets: the age of the retired scrapped asset is separated from the retired scrapped date by the time of the commissioning date, and the age of the in-service asset is separated from the current date by the time of the commissioning date; step 1.3, calculating the survival probability of various assets at each age by using a K-M method: The K-M survival probability curve is a continuous stepped curve drawn by taking the equipment service life as a horizontal axis and the survival probability as a vertical axis, and is used for describing the relation between the equipment service life and the survival probability, the survival probability curve is calculated by using the rejection ratio, and a hypothesis is given before calculation, wherein when the equipment service life is 0 year, the survival probability is 1, and the iterative calculation formula of the survival probability curve is as follows: Wherein Is the first The time points, i.e. the ages, Is at The number of assets discarded at the time of age, Is at The number of assets still in service at age; step 1.4, calculating a reference scrapping probability time sequence: After the survival function is obtained, further calculating the survival difference between adjacent age groups, wherein the calculation formula of the scrapping probability curve is as follows: 。
3. the grid asset wall risk prediction algorithm based on the KMP method according to claim 1, wherein the environmental pressure index time sequence calculation step of the step 2 is as follows: step 2.1, firstly, collecting external factor historical data, wherein the index used by environmental impact in the external factor historical data is the corrosion coefficient of each age group of the asset, the index used in the running state is the annual accumulated failure times of the asset, and the index used in the operation and maintenance strategy is the annual operation and maintenance times of each age group; then, the external factor historical data is formatted into a ds format required by a Prophet model, outliers are removed through a Z-score method, and logarithmic transformation is carried out, wherein the prediction structure of the Prophet model is expressed as follows: , wherein, Is that Time series values of time of day; as trend terms, for modeling non-periodic changes in time series; is a periodic term representing a periodic variation in the time series; for a specific event item, representing the external regression quantity, namely the abnormal influence of the specific event on the time sequence; As an error term, it is generally assumed that it obeys a normal distribution; Step 2.2, setting trend items, and marking specific events: Setting a trend item as accumulated rejection probability, marking a specific event, and adding corrosion coefficients, fault times and operation and maintenance times as features by using an add_regress function of a Prophet model; Step 2.3, training the propset model: Adding the external factor historical data by an add_regress method, and learning how to pull up or pull down the reference probability by the Prophet model; step 2.4, judging the rationality of the model error: Calculating the Root Mean Square Error (RMSE) of the Prophet model, and if the RMSE is more than 0.10, re-acquiring data or checking abnormal values until the error is reasonable; step 2.5, drawing a prediction graph for visual evaluation: Visual evaluation of the performance of a Prophet model on a power grid asset rejection probability curve is carried out by adopting the drawing function of the Prophet for calendar time And (3) carrying out the graphics, remapping virtual 'dates' used in the Prophet back to the 'ages' of the assets, and resetting the display reference curve, the model fitting curve and the scene analysis curves under different external environments.
4. The grid asset wall risk prediction algorithm based on the KMP method according to claim 1, wherein the asset scrapping risk after dynamic adjustment in the step 3 is calculated by the following method: step 3.1, calculating an adjustment factor: various asset scrapping amount results predicted by Prophet method Expected scrap amount from reference Comparing to obtain an adjustment factor For the purpose of Year adjustment factor: ; step 3.2, calculating the scrapped probability after adjustment: For a future year All in-service assets Calculating the adjusted rejection probability: wherein From K-M being an asset At the position of Obtaining the annual standard rejection probability, and obtaining the adjusted rejection probability; step 3.3, correcting various rejection probability curves: and correcting the reference scrapping probability curve according to the adjusted scrapping probability to finish dynamic adjustment.
5. The grid asset wall risk prediction algorithm based on the KMP method according to claim 1, wherein the specific implementation steps of the step 4 are as follows: step 4.1, setting an asset wall admission threshold value: Defining asset wall admission thresholds , The threshold reflects the upper limit of tolerance of the power grid enterprise to asset risks, and according to the power grid management requirement, setting: when the predicted accumulated rejection rate of a certain type of asset exceeds 80%, the corresponding age is the 'in-wall age' of the type of asset; step 4.2, calculating a dynamic accumulated rejection probability curve: based on the dynamically adjusted asset rejection probability obtained in step 3 I.e. in a specific year At a different age The scrapping probability density of the asset in the current year is converted into accumulated scrapping probability CDF; for a particular asset class Assuming a maximum design life or a statistically observed maximum age of Then the class of asset is at age Dynamic cumulative probability of rejection at the time The calculation formula is as follows: , wherein, Representing the age of the asset, ; Step 4.3, judging the in-wall years of various assets: Traversing age of various assets Searching minimum age meeting cumulative probability threshold condition, and wall-in years of various assets The definition is as follows: ; step 4.4, generating an asset in-wall judgment standard table: according to the calculation, a power grid asset in-wall year judgment standard table aiming at the target predicted year is generated and used as the basis for subsequent risk calculation.
6. The grid asset wall risk prediction algorithm based on the KMP method according to claim 1, wherein the specific implementation steps of the step 5 are as follows: Step 5.1, setting a predicted target year and calculating the future service life of the asset: first, the target year to be predicted is specified Then, traversing the current in-service asset list of the power grid, and calculating the estimated service age of the power grid in the target year based on the operation date of the asset, wherein the calculated formula is as follows: Wherein, the method comprises the steps of, Representing the first in the manifest The asset of the item, The year of delivery of the asset; Step 5.2, screening a wall-in asset list: Comparing the calculated estimated service life with the dynamic in-wall years of the various assets determined in the step 4, screening out all risk assets which will enter the asset wall in the target year, setting the first The category to which the item asset belongs is The dynamic in-wall years of the category are The decision logic is as follows: If it is Judging the asset as a wall-in asset and incorporating the wall-in asset into a risk calculation range, otherwise, judging the asset as a safety asset and not counting the risk scale; Step 5.3, measuring and calculating the total risk scale of the asset wall: accumulating the value of all the screened wall-in assets to obtain the target year Risk total scale of grid asset wall The calculation formula is as follows: , wherein, Asset collection meeting the in-wall condition for the target year; for the asset wall size sought.
7. The grid asset wall risk prediction algorithm based on the KMP method according to claim 1, wherein the specific implementation steps of the step 6 are as follows: Step 6.1, predicting future investment capacity of the power grid: collecting fixed asset investment completion data of the power grid in the past N years, and carrying out a compound annual average growth rate formula: wherein For the last year of investment amount, Is that Initial investment amount before the year; Then forecast the investment capacity of the target year based on the investment amount of the last year Calculating the future target year Is of the power grid investment capacity The calculation formula is as follows: wherein The time interval from the current year to the target year is given in years; step 6.2, calculating the capital requirement of the asset wall reconstruction, and considering the capital time value: asset wall size measured in step 5 Calculated based on the 'original value of the asset', and then the original value is corrected by introducing interest or cost increase coefficient to obtain the actual requirement of the modification of the asset wall The calculation formula is Wherein For a preset annual average cost increase rate or interest rate, referencing the object price index CPI; a time span for the asset to be delivered to the target year; step 6.3, calculating a risk ratio: comparing the corrected asset wall transformation requirement with the predicted power grid investment capacity, and calculating the asset wall risk ratio The calculation formula is as follows: ; step 6.4, determining the risk level of the asset wall: according to the actual requirement of the power grid asset management, dividing risks into three levels according to the size of the ratio R, and guiding the establishment of investment strategies: low risk The investment capacity of the power grid is greater than or equal to the requirement of the reconstruction of the asset wall, the current investment plan of the enterprise covers the upcoming requirement of the scrapping update of the asset, the fund flow is healthy, and the pressure of eliminating the wall is avoided; Risk in The asset wall transformation requirement exceeds the investment capacity, but the exceeding amplitude is within 20%; high risk The requirement of the modification of the asset wall is far beyond the investment capacity, and the gap amplitude exceeds 20%.
8. The KMP method-based power grid asset wall risk prediction system is characterized by comprising a reference probability measurement module, an environmental pressure prediction module, a risk dynamic correction module, a wall-in age judgment module, a risk scale measurement module and a risk level assessment module, wherein the reference probability measurement module is used for collecting a power grid asset historical retired scrapped asset list and a current in-service asset list, measuring and calculating a reference scrapped amount according to the age of the power grid retired scrapped asset, constructing a reference scrapped probability time sequence, the environmental pressure prediction module is used for acquiring external factor historical data, namely environmental influence, running state and operation and maintenance strategy historical data, predicting the change trend of future external influence factors by using a time sequence model, generating an environmental pressure index time sequence, the risk dynamic correction module is used for correcting the reference scrapped probability time sequence by using the environmental pressure index time sequence, generating dynamic adjusted asset scrapped risk probabilities, the wall-in age judgment module is used for determining the risk of various types of assets by combining the dynamic adjusted asset scrapped risk probabilities based on a preset asset scrapped threshold, the wall-in-age judgment module is used for calculating the risk scale of all the power grid asset models, and calculating the specified risk scale of the power grid cost of the power grid asset scale, and calculating the risk scale of the specified risk scale.

Description

KMP method-based power grid asset wall risk prediction algorithm and system Technical Field The invention relates to the field of power grid asset wall risk prediction, in particular to a power grid asset wall risk prediction algorithm and system based on a KMP method. Background Asset wall models are key asset management tools, and by drawing life cycle distribution broken line area diagrams of enterprise total assets in an asset age and asset value coordinate system, wall-shaped forms are presented, and future asset aging retirement peaks can be visually predicted. When the amount of aged retired assets exceeds the investment improvement capability, an asset wall risk is created. Asset wall risk prediction provides basis for risk control and investment strategy formulation by predicting in-service asset life, locating asset wall age, and quantifying future asset wall scale. In recent decades, the construction of the power grid in China experiences several investment peaks, and the power grid assets densely put into operation in the same period are concentrated to age in the future, so that the risk of asset walls is formed. If the risk cannot be accurately predicted and the asset iteration upgrade is timely arranged, the safe and stable operation of the power grid system is threatened. Therefore, the establishment of an accurate prediction algorithm and system is a key for maintaining the long-term reliability of the power grid, and is also a realistic appeal of the asset management department of the power grid enterprise. The traditional asset wall prediction method only adopts a simple average life translation method, has coarse asset analysis granularity, does not consider the difference of the life of different types of assets, and is difficult to accurately predict the scale of the asset wall. In order to improve the prediction accuracy, domestic scholars develop related researches that Li Zhiwei and other methods for predicting the value of the asset wall model based on the value chain management theory are provided, liu Yihe and other methods are combined with the expected service life to construct a future asset wall value prediction model, and partial scholars introduce a Kaplan-Meier (K-M) method into the life prediction of the power equipment to provide a new thought for improving the asset wall prediction. The K-M method originates from survival analysis in the field of statistics, is proposed by Edward L, kaplan and Paul Meier in 1958 and is used for processing deleted data and estimating survival probability, and the core is to construct a survival curve through non-parameter estimation, so that the K-M method is suitable for reliability analysis scenes such as equipment life, disease prognosis and the like. In the field of engineering in China, the K-M method is applied to the residual life prediction of a satellite attitude control system in Qi Haiming and 2021, lei Tong is used for performance degradation analysis of an emergency lamp battery assembly in 2021, the effectiveness of the emergency lamp battery assembly in reliability evaluation of power equipment is verified, the K-M method is used in asset wall prediction for the first time in 2023 in Wangkai, and Jiang Changtai proposes that the K-M method is combined with bearing fault data to characterize the service life of equipment in 2024. However, the K-M method has the limitation in the risk analysis of the asset walls of the power grid that the result belongs to static estimation, and the future scrapping mode is assumed to be consistent with the history rule, so that the dynamic influence of external factors such as environmental influence, running state, operation and maintenance strategy and the like on the life of the asset is difficult to reflect, and the prediction precision is difficult to meet the fine requirement of the asset management of the power grid. Disclosure of Invention The invention aims to provide a grid asset wall risk prediction algorithm and a system based on a KMP method, so as to solve the problems in the background technology. In order to achieve the purpose, the invention provides a power grid asset wall risk prediction algorithm based on a KMP method, which comprises the following steps: Step 1, collecting a historical retired scrapped asset list and a current in-service asset list of a power grid, and calculating a reference scrapped amount according to the age of the retired scrapped asset of the power grid to obtain a reference scrapped probability time sequence; Step 2, collecting external factor historical data, wherein the external factor historical data comprises, but is not limited to, environmental influence, running state and operation and maintenance strategy, predicting external influence factors which fluctuate with time and influence the reference rejection probability, and obtaining an environmental pressure index time sequence; step 3, correcting the reference scrapping probability time