CN105787606A - Power dispatching online trend early warning system based on ultra short term load prediction - Google Patents

Abstract

The present invention relates to the technical field of applying online safety analysis results to load forecasting, and in particular to an online trend warning system for power dispatching based on ultra-short-term load forecasting. Specifically, the future of the power grid is calculated by counting, analyzing and forecasting load data. To form a short-term trend warning of the power grid and a post-accident treatment plan. The invention includes the following operation steps: short-term load forecasting based on big data analysis of power consumption; power flow trend analysis based on ultra-short-term load obtained by ARIMA algorithm; power grid risk assessment and alarm; equipment maintenance progress management; data access panoramic monitoring; Deal with the accident; restore the power supply after the accident. The invention can realize the linkage of online trend early warning and scheduling plan change information, and constitute a provincial and local integrated trend analysis early warning and auxiliary decision-making system.

Description

An online trend warning system for power dispatching based on ultra-short-term load forecasting

technical field

The present invention relates to the technical field of applying online safety analysis results to load forecasting, and in particular to an online trend warning system for power dispatching based on ultra-short-term load forecasting. Specifically, it calculates the future of the power grid through statistics, analysis and forecasting of load data. To form a short-term trend warning of the power grid and a post-accident treatment plan.

Background technique

At present, the safety and stability analysis of power grids in various provinces is gradually developing from offline to online. The online safety analysis system based on D5000 platform including static, transient, dynamic, voltage, frequency, small disturbance stability calculation and short-circuit current calculation has been widely used. However, when the power grid is in an abnormal state, it still needs to rely on human experience for judgment and processing, and dispatchers are prone to cognitive impairment when they need to process massive amounts of information in a short period of time. The development of power grid dispatching cannot get rid of the negative influence of factors considered by dispatchers, and the degree of intelligence is not high.

In order to reduce the interference of human factors in intelligent dispatching and improve dispatching timeliness and rapid response capability, power dispatching trend perception technology has emerged as the times require. On the basis of the existing online analysis and early warning technology, the early warning function for the current operating status is extended to the ultra-short term, which effectively fills the gap in the trend analysis function between the online early warning and the 96-point safety check before the day, and helps the dispatching and operating personnel to grasp the future period The development trend of the power grid within a certain period of time should be prepared, and a comprehensive and comprehensive evaluation of security, stability and multi-time scales can be fully realized.

Contents of the invention

Aiming at the deficiencies in the above-mentioned prior art, the present invention proposes an online trend warning system for electric power dispatching based on ultra-short-term load forecasting. Power grid short-term trend warning and post-accident handling plan.

In order to achieve the above-mentioned purpose of the invention, the present invention is achieved through the following technical solutions:

An online trend warning system for power dispatching based on ultra-short-term load forecasting, comprising the following steps:

(1) Short-term load forecasting based on big data analysis of electricity consumption;

(2) Perform power trend analysis based on the ultra-short-term load obtained by the ARIMA algorithm;

(3) Power grid risk assessment alarm;

(4) Equipment maintenance schedule management;

(5) Panoramic monitoring of data access;

(6) Dealing with accidents;

(7) Restoring power supply after the accident.

The short-term load forecasting based on big data analysis of electricity consumption includes: using time series method for short-term load forecasting; time series method is a quantitative forecasting method, which is widely used as a common forecasting method in data mining; There are two tasks in modeling time series, one is to analyze how the data of the current period is affected by the data of previous periods, and the other is the regularity of variables in time changes.

The time series algorithm described is an ARIMA model, and the full name of the ARIMA model is the AutoRegressive Integrated Moving Average Model (AutoRegressiveIntegratedMovingAverageModel, abbreviated as ARIMA), which is a well-known time series prediction proposed by Box and Jenkins in the early 1970s. method, so it is also called Box-Jenkins model, Box-Jenkins method; ARIMA (p, d, q) is called differential autoregressive moving average model, AR is autoregressive, p is autoregressive item; MA is moving average , q is the number of moving average items, and d is the number of differences made when the time series becomes stationary; the basic idea of the ARIMA model is: regard the data sequence formed by the forecast object over time as a random sequence, and use a certain mathematical model To approximately describe this sequence; once this model is identified, it can predict future values from the past and present values of the time series; modern statistical methods and econometric models have been able to help companies predict the future to some extent;

The ARIMA model refers to the model established by converting the non-stationary time series into a stationary time series, and then regressing the dependent variable only on its lag value and the present value and lag value of the random error item; in the identification process of the ARIMA model , two tools are mainly used: autocorrelation function, referred to as ACF; partial autocorrelation function, referred to as PACF; and their respective correlation diagrams, that is, ACF and PACF are plotted against the lag length; The k-order autocorrelation coefficient (r _K ) is defined as its k-order autocovariance divided by its variance;

${\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>\frac{\underset{\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; k</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; the\; y</span>}}_{\mathrm{<span\; class="notranslate">\; t</span>}}<span\; class="notranslate">\; -</span>\stackrel{<span\; class="notranslate">\; \&OverBar;</span>}{\mathrm{<span\; class="notranslate">\; the\; y</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; the\; y</span>}}_{\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>\stackrel{<span\; class="notranslate">\; \&OverBar;</span>}{\mathrm{<span\; class="notranslate">\; the\; y</span>}}<span\; class="notranslate">\; )</span>}{{<span\; class="notranslate">\; \&Sigma;</span>}_{\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}^{\mathrm{<span\; class="notranslate">\; no</span>}}{<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; the\; y</span>}}_{\mathrm{<span\; class="notranslate">\; t</span>}}<span\; class="notranslate">\; -</span>\stackrel{<span\; class="notranslate">\; \&OverBar;</span>}{\mathrm{<span\; class="notranslate">\; the\; y</span>}}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ;</span>$

In the formula: y is a sequence; r _K is the k-order autocorrelation coefficient; n represents the upper bound, and t represents the lower bound;

Equation (1) is a function of k, also known as the autocorrelation function ACF(k); the partial autocorrelation function PACF(k) measures the correlation between the two lag variables after eliminating the influence of the intermediate lag item;

The ARIMA(p,d,q) model is the ARMA(p,q) model after the d-order differential transformation, and the general form of the ARMA(p,q) model is formula (2):

y _t = c+φ ₁ y _t-1 +...+φ _p y _tp +ε _t +θ ₁ ε _t-1 +...+θ _q ε _tq (2);

In the formula: c is noise mean value; φ ₁ , φ ₂ ... φ _p are autoregressive parameters; θ ₁ , θ ₂ ... θ _q are moving average parameters; ε _t , ε _t-1 ... ε _tq is a white noise process;

Therefore, the ARIMA model can be used to predict the ultra-short-term load of the power grid, and then carry out the next trend analysis.

The power flow trend analysis based on the ultra-short-term load obtained by the ARIMA algorithm includes: the online trend analysis early warning system calculates the future state power flow based on the next 15 minutes ultra-short-term load forecast, ultra-short-term wind power and photovoltaic forecast and real-time power generation plan, and then uses The hardware and software environment of existing DSA applications perform trend analysis; first calculate the current section and then calculate the future section. After the online data is integrated, the calculation of the current section is started, and the trend data inspection and future state and power flow calculation are performed at the same time; after the calculation of the current section is completed, Use the existing machine group and calculation function to calculate the cross-section in the next 15 minutes and analyze the evolution trend of the power grid; at the same time, through real-time collection and comparison of equipment power flow changes, monitor equipment power flow mutations and perceive various power grid disturbances in real time; through large screen Display the sudden change of a certain equipment before and after the time and the change rate at the same time today and yesterday to help the dispatcher analyze the main reason for the change of the power flow of a certain component.

The trend analysis function module is located in the online security and stability analysis. First, the ultra-short-term load forecast obtained by the ARIMA algorithm is input as the forecasting module in the scheduling application; The planning and power generation planning modules are pushed into the future power flow calculation function module in the online security and stability analysis; at the same time, the future power flow calculation function module also comprehensively considers the online integration flow from the data integration function module and the DC line plan for short-term transaction management Comprehensive analysis of inter-provincial contact line plan and other data; next, the future power flow calculation function module outputs the trend flow as the input of the trend analysis function module; finally, the trend analysis result of the trend analysis function module is input to the grid real-time monitoring and intelligent alarm , operation analysis and evaluation, and scheduling operation auxiliary decision-making applications for display and early warning, and completed the whole process of calculation, operation and output of trend analysis.

The power grid risk assessment alarm includes: introducing power grid flow, equipment maintenance, real-time weather, online security analysis, power generation and supply balance, etc. Layered and graded artificial neural network model and system; the risk assessment process of power grid operation includes information collection, risk identification, risk assessment and grading, and risk early warning; the information collection function is to collect and process the "risk identification" and "risk assessment" links All kinds of information needed; risk identification refers to the classification and quantification of various risk factors and risk indicators that affect the safe operation of the power grid, and the establishment of rating scores with percentages as indicators.

The maintenance progress management of the equipment includes: realizing the display and query of various voltage levels and various maintenance equipment respectively; using the time progress Gantt chart to display the maintenance progress in real time; using the actual switch status access method of the equipment to grasp the equipment status in real time; automatically obtaining the power grid Current control temporary section.

The data access panoramic monitoring includes: the system accesses a variety of different data types of business systems, the interface types are rich, the amount of information is large, data screening and optimization is realized, and there is room for further expansion; the system accesses telemetry, telemetry There are 180,000 information points, and more than 1,200 telemetry points are calculated each time the trend changes. According to the system refresh every 15 seconds, 76,778 dynamic telemetry points are calculated, and the calculated and stored data reaches 450 million pieces every day.

The handling during the accident includes: when an accident occurs in the power grid, in order to ensure the stability of the power grid, the control personnel need to consult a large amount of first-off information as a reference for system stability, such as reviewing procedure information, plan information, load conditions, line parameters, and accident handling history information, etc.; however, the amount of information is large and there is no relationship between them, and it takes time and effort to inquire; for this reason, the relevant information is collected through the platform uniformly, and the relationship between the information is established; when an accident occurs in the power grid, according to the content of the accident information, the platform The analysis and query results of accident-related information can be directly pushed in front of the control personnel, providing data support for the control personnel to deal with the accident.

The restoration of power supply after the accident includes: in the past, the restoration of power supply required the control personnel to report one by one by telephone; after the online trend analysis and early warning system platform was established, the dispatcher could issue an operation order to restore power supply through the platform. Orders can be directly referenced through past incident handling history, reducing overall recovery time.

Advantageous effect of the present invention is:

my country's current power grid data application stays at the basic level, and there are few other advanced applications except for daily business management. The present invention proposes to apply online security analysis results in the field of load forecasting, integrate main network online security analysis with applications such as emergency command platform, intelligent pre-plan system, and power grid risk assessment, to form short-term trend warning and post-accident treatment plans for power grids. Realized the linkage of online trend early warning and dispatch plan change information, forming a provincial-regional integrated trend analysis early warning and auxiliary decision-making system.

The basis of short-term load forecasting is the big data analysis technology of user electricity consumption. Through the historical power consumption of a certain area, analyze the load characteristics, and establish the time model of the load. By monitoring the big data changes of operating electricity consumption, the power company senses the sudden change of regional load, and then feeds the load forecast information to the front-line dispatching department, thereby triggering the online trend calculation of the dispatching department, effectively warning before the accident, and shortening the power outage and emergency repair after the accident time.

Description of drawings

Fig. 1 is a system block diagram of the present invention;

Figure 2 is the power grid risk assessment model.

detailed description

The present invention is an online trend warning system for power dispatching based on ultra-short-term load forecasting, which specifically includes the following steps:

1. Short-term load forecasting based on big data analysis of electricity consumption.

Short-term load forecasting using time series methods. Time series method is a quantitative forecasting method, which is widely used as a common forecasting method in data mining. There are two tasks in modeling time series, one is to analyze how the data of the current period is affected by the data of previous periods, and the other is the regularity of variables in time changes. The time series algorithm selected by the present invention is the ARIMA model. The full name of the ARIMA model is the AutoRegressive Integrated Moving Average Model (ARIMA for short). It is a well-known time series prediction method proposed by Box and Jenkins in the early 1970s, so it is also called the Box-Jenkins model. , Box-Jenkins method. Among them, ARIMA (p, d, q) is called the differential autoregressive moving average model, AR is autoregressive, p is the autoregressive item; MA is the moving average, q is the number of moving average items, and d is the time series when it becomes stable. the number of differences. The basic idea of the ARIMA model is to regard the data sequence formed by the forecast object over time as a random sequence, and use a certain mathematical model to approximate this sequence. This model, once identified, can predict future values from the past and present values of the time series. Modern statistical methods and econometric models have been able to help companies predict the future to some extent.

The ARIMA model refers to the model established by converting the non-stationary time series into a stationary time series, and then regressing the dependent variable only on its lag value and the present value and lag value of the random error item. In the identification process of the ARIMA model, two tools are mainly used: autocorrelation function, referred to as ACF; partial autocorrelation function, referred to as PACF; and their respective correlation diagrams, that is, ACF and PACF are plotted against the lag length. For a sequence y, its k-th order autocorrelation coefficient (r _K ) is defined as its k-th order autocovariance divided by its variance.

${\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>\frac{\underset{\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; k</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; the\; y</span>}}_{\mathrm{<span\; class="notranslate">\; t</span>}}<span\; class="notranslate">\; -</span>\stackrel{<span\; class="notranslate">\; \&OverBar;</span>}{\mathrm{<span\; class="notranslate">\; the\; y</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; the\; y</span>}}_{\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>\stackrel{<span\; class="notranslate">\; \&OverBar;</span>}{\mathrm{<span\; class="notranslate">\; the\; y</span>}}<span\; class="notranslate">\; )</span>}{{<span\; class="notranslate">\; \&Sigma;</span>}_{\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}^{\mathrm{<span\; class="notranslate">\; no</span>}}{<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; the\; y</span>}}_{\mathrm{<span\; class="notranslate">\; t</span>}}<span\; class="notranslate">\; -</span>\stackrel{<span\; class="notranslate">\; \&OverBar;</span>}{\mathrm{<span\; class="notranslate">\; the\; y</span>}}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ;</span>$

In the formula: y is the sequence; r _K is the k-order autocorrelation coefficient. n represents the upper bound, and t represents the lower bound.

Equation (1) is a function of k, also known as the autocorrelation function ACF (k). Partial autocorrelation function PACF(k) measures the correlation between two lag variables after eliminating the influence of intermediate lag items.

The ARIMA(p,d,q) model is the ARMA(p,q) model after the d-order differential transformation, and the general form of the ARMA(p,q) model is formula (2):

y _t = c+φ ₁ y _t-1 +...+φ _p y _tp +ε _t +θ ₁ ε _t-1 +...+θ _q ε _tq (2);

In the formula: c is noise mean value; φ ₁ , φ ₂ ... φ _p are autoregressive parameters; θ ₁ , θ ₂ ... θ _q are moving average parameters; ε _t , ε _t-1 ... ε _tq is a white noise process.

Therefore, the ARIMA model can be used to predict the ultra-short-term load of the power grid, and then carry out the next trend analysis.

2. Perform trend analysis based on the ultra-short-term load forecast obtained by the ARIMA algorithm.

The online trend analysis early warning system is based on the next 15 minutes ultra-short-term load forecast, ultra-short-term wind power and photovoltaic forecast, and real-time power generation plan to calculate the future state and power flow, and then use the existing DSA application hardware and software environment for trend analysis. Calculate the current section first and then calculate the future section. After the online data is integrated, the calculation of the current section is started, and the trend data inspection and future state and power flow calculation are performed at the same time. After the calculation of the current section is completed, the calculation of the section in the next 15 minutes is carried out by using the existing machine group and the calculation function, and the evolution trend of the power grid is analyzed. At the same time, through real-time collection and comparison of equipment power flow changes, it monitors equipment power flow mutations and senses various power grid disturbances in real time. By displaying the sudden change of a certain equipment and the change rate at the same time today and yesterday on the large screen, it helps the dispatcher to analyze the main reason for the change of the power flow of a certain component.

As shown in Figure 1, the trend analysis function module is located in the online security and stability analysis. First, the ultra-short-term load forecast obtained by the ARIMA algorithm is used as the input of the forecasting module in the scheduling application. Then, push the prediction module into the future power flow calculation function module in the online security and stability analysis together with the maintenance plan and power generation plan modules in the dispatch planning application. At the same time, the future trend calculation function module also comprehensively considers the online integration trend from the data integration function module, as well as the DC line plan and inter-provincial contact line plan for short-term transaction management for comprehensive analysis. Next, the future power flow calculation function module outputs the trend power flow as the input of the trend analysis function module. Finally, the trend analysis results of the trend analysis function module are input to the grid real-time monitoring and intelligent alarm, operation analysis and evaluation, and scheduling operation auxiliary decision-making applications for display and early warning, completing the whole process of calculation, operation and output of trend analysis.

3. Power grid risk assessment and alarm.

Introduce power grid flow, equipment maintenance, real-time weather, online security analysis, power generation and supply balance, etc. into the risk assessment system, through continuous in-depth analysis of the data of the provincial power grid, and classify, layer, and grade artificial neural network models according to risk influencing factors and system. As shown in Figure 2, the power grid operation risk assessment process includes information collection, risk identification, risk assessment, risk grading, and risk early warning. The information collection function is to collect and process all kinds of information required in the "risk identification" and "risk assessment" links. Risk identification refers to the classification and quantification of various risk factors and risk indicators that affect the safe operation of the power grid, and the establishment of rating scores with percentages as indicators.

4. Equipment maintenance schedule management.

Realize the display and query of various voltage levels and various types of maintenance equipment; use the time progress Gantt chart to display the maintenance progress in real time; use the actual switch status access method of the equipment to grasp the equipment status in real time; automatically obtain the temporary section of the power flow control.

5. Data access panoramic monitoring

The system is connected to a variety of business systems with different data types, with rich interface types and a large amount of information, which realizes the optimization of data screening and has room for continuous expansion. The system has access to 180,000 telemetry and remote signaling points, and more than 1,200 telemetry points are calculated each time the power flow changes. According to the system refresh every 15 seconds, 76,778 dynamic telemetry points are calculated, and the calculated and stored data reaches 450 million pieces every day.

6. Deal with accidents.

When an accident occurs in the power grid, in order to ensure the stability of the power grid, regulators need to consult a large amount of pre-official information as a reference for system stability, such as reviewing regulatory information, plan information, load conditions, line parameters, and accident handling history information. However, the amount of information is large and there is no relationship between each other, and the query is time-consuming and laborious. To this end, the relevant information is collected uniformly through the platform, and the association relationship between the information is established. When an accident occurs in the power grid, according to the content of the accident information, the platform can directly push the analysis and query results of the accident-related information to the regulators, providing data support for the regulators to deal with the accident.

7. Restore power supply after the accident.

In the past, the process of restoring power supply required the regulators to report one by one by phone. After the online trend analysis and early warning system platform is established, dispatchers can issue operation orders to restore power supply through the platform. Most operation orders to restore power supply can be directly quoted from the past accident handling history, shortening the overall recovery time.

Claims (9)

1. based on the online tendency early warning system of the power scheduling of ultra-short term, it is characterized in that: include following operating procedure:

(1) based on the short-term load forecasting of the big data analysis of power consumption；

(2) super short period load drawn based on ARIMA algorithm carries out trend trend analysis；

(3) power grid risk assessment alarm；

(4) overhaul of the equipments progress control；

(5) data access panoramaization monitors；

(6) accident processes；

(7) restore electricity after accident.

2. a kind of online tendency early warning system of the power scheduling based on ultra-short term according to claim 1, is characterized in that: the described short-term load forecasting based on the big data analysis of power consumption, including:

Time Series Method is used to carry out short-term load forecasting；Time series method is a kind of quantitative forecasting technique, is widely used as a kind of conventional predicting means in data mining；Two tasks to time series modeling, one is analyze how current-period data is subject to the data influence of former phases, and two is variable regularity on the time changes.

3. a kind of online tendency early warning system of the power scheduling based on ultra-short term according to claim 2, it is characterized in that: described time series algorithm is ARIMA model, ARIMA model full name is ARMA model (AutoRegressiveIntegratedMovingAverageModel, brief note ARIMA), it is the famous Time Series Forecasting Methods proposed the beginning of the seventies by Bock think of (Box) and Charles Jenkins (Jenkins), so being also called Box-Jenkins model, Bock think of-Jenkins method；Wherein (p, d, q) be called difference ARMA model to ARIMA, and AR is autoregression, and p is autoregression item；MA is rolling average, and q is rolling average item number, and d is the difference number of times that time series is done when becoming steady；The basic thought of ARIMA model is: the data sequence formed predicting object to elapse in time is considered as a random sequence, carrys out this sequence of approximate description with certain mathematical model；This model just can predict future value from seasonal effect in time series past value and present value after identified；Modern Statistical Methods, econometric model have been able to help enterprise that future is predicted to a certain extent；

ARIMA model, refers to and nonstationary time series is converted into stationary time series, then only to its lagged value and the present worth of stochastic error and lagged value, dependent variable is returned the model set up；In the identification process of ARIMA model, mainly use two instruments: auto-correlation function, be called for short ACF；Partial autocorrelation function, is called for short PACF；And the relevant figure of each of which, namely ACF, PACF trace designs relative to the length of lag；For a sequences y, its kth rank autocorrelation coefficient (r_K) it is defined as its k rank auto-covariance variance divided by it；

$r_k=\frac{\underset{t=1}{\overset{n-k}{\&Sigma;}}(y_t-\stackrel{\&OverBar;}{y})(y_{t+k}-\stackrel{\&OverBar;}{y})}{{\mathrm{\&Sigma;}}_{t=1}^n{(y_t-\stackrel{\&OverBar;}{y})}^2}---\left(1\right);$

In formula: y is sequence；r_KFor the autocorrelation coefficient of k rank；N represents the upper bound, and t represents lower bound；

Formula (1) is the function about k, also referred to as auto-correlation function ACF (k)；Partial autocorrelation function PACF (k) has measured the dependency relation after delayed item in the middle of elimination affects between two lagged variables；

ARIMA (p, d, q) model be ARMA after the differential transformation of d rank (p, q) model, ARMA (p, q) general type of model is formula (2):

y_t=c+ φ₁y_t-1+...+φ_py_t-p+ε_t+θ₁ε_t-1+...+θ_qε_t-q(2)；

In formula: c is noise average；φ₁,φ₂...φ_pFor auto-regressive parameter；θ₁, θ₂...θ_qFor rolling average parameter；ε_t,ε_t-1...ε_t-qFor white-noise process；

Therefore, electrical network super short period load can be predicted by ARIMA model, and then carry out next step trend analysis.

4. a kind of online tendency early warning system of the power scheduling based on ultra-short term according to claim 1, is characterized in that: the described super short period load drawn based on ARIMA algorithm carries out trend trend analysis, including:

Online trend analysis early warning system carries out future-state Load flow calculation based on the prediction of lower 15 points of ultra-short term, ultra-short term wind-powered electricity generation and photovoltaic and real-time generation schedule, then utilizes the existing DSA hardware and software environment applied to carry out trend analysis；First calculate current section and calculate following section again, after online data is integrated, start the calculating of current section, carry out trend data inspection and future-state Load flow calculation simultaneously；After current section has calculated, utilize an existing group of planes and computing function to carry out the calculating of following 15 minutes sections, analyze electrical network evolving trend；Meanwhile, by i.e. Real-time Collection, comparative apparatus power flow changing, carry out the sudden change of supervision equipment trend, all kinds of grid disturbance of real-time perception；By show on giant-screen before and after a certain equipment moment catastrophe and modern yesterday synchronization rate of change, help dispatcher analyzes the change main cause of the trend of a certain element；

Described trend analysis functional module is arranged in on-line security and stability analysis, and first, the ultra-short term drawn by ARIMA algorithm inputs as the prediction module in the application of operation plan class；Then, prediction module coordinate the repair schedule in the application of operation plan class, generation schedule module are pushed in the following Load flow calculation functional module in on-line security and stability analysis；Meanwhile, following Load flow calculation functional module has also considered the online tideway integrating from data integration function module, and AC line plan and inter-provincial interconnection plan etc. the data of short-term trading management are comprehensively analyzed；It follows that following Load flow calculation functional module output trend trend is as the input of trend analysis functional module；Finally, the trend analysis result input of trend analysis functional module is shown and early warning in applying to electrical network monitor in real time with intelligent alarm, operating analysis and evaluation, management and running aid decision, completes the calculating of trend analysis, operation and output overall process.

5. a kind of online tendency early warning system of the power scheduling based on ultra-short term according to claim 1, is characterized in that: described power grid risk assessment alerts, including:

By electric network swim, overhaul of the equipments, real-time weather, safety on line analysis, send out for introducing risk evaluating systems such as electric equilibrium, by constantly analysing in depth the data of electrical network inside the province, undertaken classifying by venture influence factor, be layered, the artificial nerve network model of classification and system；Operation of power networks risk assessment flow process includes information gathering, Risk Identification, risk assessment and deciding grade and level, Risk-warning；Information collection function is for collecting and process the various information that " Risk Identification " needs with " risk assessment " link；Namely the various risks factor affecting electric power netting safe running, risk indicator are carried out classification quantitative by Risk Identification, set up the rating score being index with percentage ratio.

6. a kind of online tendency early warning system of the power scheduling based on ultra-short term according to claim 1, is characterized in that: described overhaul of the equipments progress control, including:

Realizing each electric pressure, all kinds of repair apparatus are shown respectively and inquire about；Utilize time schedule Gantt chart real-time exhibition repair schedule；Employing equipment actual switch state access way grasps equipment state in real time；Automatically obtain electric network swim and control temporary section.

7. a kind of online tendency early warning system of the power scheduling based on ultra-short term according to claim 1, is characterized in that: described data access panoramaization monitors, including:

The operation system of the multiple different types of data of system access, interface type enriches, and contains much information, it is achieved that data screening optimization, and has the space continuing extension；System access remote measurement, remote signalling point 180,000, trend sudden change calculates telemetry station more than 1200 every time, within every 15 seconds, refreshes once according to system, and 76778 dynamic telemetry points calculate, and calculates storage data every day and reaches 4.5 hundred million.

8. a kind of online tendency early warning system of the power scheduling based on ultra-short term according to claim 1, is characterized in that: process in described accident, including:

When having an accident in electrical network, for ensureing the stabilization of power grids, regulation and control personnel need to consult a large amount of information of first closing as system stability reference, such as consult protocol information, prediction scheme information, load condition, line parameter circuit value, accident treatment historical information etc.；But these contain much information, do not have incidence relation each other, and inquiry is wasted time and energy；Collect these relevant informations for this by platform is unified, and set up the incidence relation between information；When electrical network has an accident, according to accident information content, the analysis Query Result of accident relevant information can directly be pushed in face of regulation and control personnel by platform, provides data supporting for regulation and control personnel to the process of accident.

9. a kind of online tendency early warning system of the power scheduling based on ultra-short term according to claim 1, is characterized in that: restore electricity after described accident, including:

The process of restoring electricity needed regulation and control personnel to be reported one by one by the form of phone in the past；After online trend analysis early warning system platform is set up, dispatcher can pass through platform and issue the operational order restored electricity, and the operational order that great majority restore electricity by directly quoting in conventional accident treatment history, can shorten overall recovery time.