Method for controlling a technical system

Abstract

Method for controlling a technical system including a power grid, connected to at least one energy supply system. The method ensures that the consumption of electrical energy of the grid is within a given load curve by avoiding or compensating peak loads; including the steps of: a) providing data of the total load occurring in the grid to a load data processor; b) the processor processing said data and determining curves of a plurality of load components of the total load; c) determining future load curves for each component for a time period ahead; d) superimposing the future load curves and determining a future total load curve for the future total load and future peak loads; and e) controlling the technical and/or energy supply systems based on the determined future total load curve to avoid exceeding given load limits and future peak loads or allocate energy required in the future.

Claims

1. A method for controlling a technical system that includes a power grid which is connected to at least one energy supply system, in such a way that the consumption of electrical energy in said power grid is kept within a given load curve by avoiding or compensating peak loads, the method comprising the steps of: providing data of the course of the total load occurring in the power grid to a load data processor; the load data processor processing said data of the course of the total load and determining curves of a plurality of load components of the total load, including processing data of the course of the total load for a single day or a plurality of days and determining a total load curve, and the load components being determined by: extracting recursively occurring signal components from the total load curve, and/or extracting periodically occurring signal components from the total load curve, and/or extracting the noise component from the total load curve; determining future load curves for each of the load components for a time period ahead, the future total load curve being determined based on the determined total load curve; superimposing the future load curves and determining a future total load curve for the future total load and determining future peak loads; and controlling the technical system and/or the at least one energy supply system based on the determined future total load curve in order to avoid exceeding a given load limit, to avoid future peak loads or to allocate energy required for compensating predicted peak loads.

2. The method according to claim 1, further comprising the steps of: determining the load components by the application of a Fourier transform to the total load curve.

3. The method according to claim 1, further comprising the steps of: extracting a curve of a basic load component from the curve of the total load.

4. The method according to claim 1, further comprising the steps of: extracting periodically occurring signal components from the curve of the total load, which signal components correspond to operating cycles of a timetable, with which the technical system is operated.

5. The method according to claim 1, further comprising the steps of: analysing data that relate to internal influences on the curve of the total load; evaluating related load changes; and correcting the future total load curve based on the evaluated load changes.

6. The method according to claim 1, further comprising the steps of: analysing data that relate to external influences on the total load; evaluating related load changes; and correcting the future total load curve based on the evaluated load changes.

7. The method according to claim 1, further comprising the steps of: determining changes in the amplitude or phase of the curves of the determined load components and correcting the future load curves accordingly.

8. The method according to claim 1, comprising the steps of: comparing the future total load curve with at least one fixed or variable threshold in order to determine future peak loads that exceed the selected threshold.

9. The method according to claim 8, further comprising the steps of: providing a threshold for one or a plurality of ranges of the future total load curve, said threshold being based on the curve of the fundamental load component including an offset.

10. The method according to claim 8, further comprising the steps of: generating first control signals with a lead time ahead of the future peak loads, which exceed the related threshold; and controlling individual units of the technical system or the at least one energy supply system with these first control signals in such a way that said future peak loads are avoided or the required energy is allocated for the future peak loads.

11. The method according to claim 10, further comprising the steps of: controlling individual units of the technical system by fully or partially deactivating or shifting the power consumption of individual units of the technical system; or transforming kinetic energy present in the individual units of the technical system into electrical energy that is supplied to the power grid.

12. The method according to claim 1, further comprising the steps of: creating a second control signal by selecting a predicted load curve of at least one of the load components; evaluating a direct component in the selected load curve for at least one control period; removing the direct component from the selected load curve for each control period; and controlling the at least one energy supply system with the second control signal.

13. The method according to claim 12, comprising the steps of: creating a second control signal by creating a sum signal by adding predicted load curves of at least two of the load components; evaluating a direct component in the sum signal for at least one control period; removing the direct component from the sum signal for each control period; and controlling the at least one energy supply system with the second control signal.

14. The method according to claim 12, further comprising the steps of: controlling with the second control signal the first energy supply system, which is part of a control path of a control loop, with which load deviations from a basic load are covered, which basic load is supplied by the second energy supply system.

15. The method according to claim 14, further comprising the steps of: providing a guide value, which corresponds to the basic load, to the control loop; superimposing the das second control signal onto the guide value, in order to obtain a corrected guide value; deriving a measurement value from the technical system, which value corresponds to the actual energy consumption; comparing the corrected guide value with said measurement value; and determining a corresponding control deviation that is forwarded to a controller, which provides an actuating variable to the controlled first energy supply system.

16. A technical system comprising: a power grid, which is connected to at least one energy supply system, the power grid including a load data processor and a software program installed in the load data processor, with which the technical system is controlled with the method according to claim 1.

17. The technical system according to claim 16, wherein the load data processor is connected via data channels to a system controller, with which the system units of the technical system are controlled.

18. The technical system according to claim 16, wherein the load data processor is connected via data channels to a system controller, with which at least one of the energy supply system can be controlled.

19. The technical system according to claim 16, wherein the load data processor is connected via data channels to a planning data processor, from which data can be downloaded.

20. The technical system according to claim 16, wherein the technical system is a railway system.

21. The method according to claim 1, wherein the technical system is a railway system.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) Below the invention is described in detail with reference to the drawings. Thereby show:

(2) FIG. 1 a course of a total load PG in a power grid BN of a technical system EA, particularly a railway system, over 24 hours exhibiting higher loads and peak loads in in the time ranges around 08:00 and 18:00;

(3) FIG. 2 the superposition PGG of the courses of the total load PG, PGG in the power grid BN of the railway system EA that were recorded for several days in the time domain and the averaged course of the total load PGG after a Fourier transform in the frequency domain;

(4) FIG. 3a load oscillations resulting from the Fourier transform of the averaged course of the total load PGG with the indication of the amplitude, the frequencies and the cycle duration for frequencies of up to 2 mHz;

(5) FIG. 3b load oscillations resulting from the Fourier transform with the indication of the amplitude, the frequencies and the cycle duration for frequencies of above 2 mHz;

(6) FIG. 4a the course of a total load PG or the averaged course of a total load PGG for 24 hours;

(7) FIG. 4b the course of the basic load PS0 over 24 hours extracted from the course of the total load PG, PGG of FIG. 4a after removing periodic signal components and noise components;

(8) FIG. 4c the course of a first load component PS1 over 24 hours extracted from the course of the total load PG, PGG of FIG. 4a, which exhibits a cycle duration of 30 minutes;

(9) FIG. 4d the envelope of a second load component PS2 over 24 hours extracted from the course of the total load PG, PGG of FIG. 4a, which exhibits the cycle duration 1 minute, and a detailed view of this load oscillation PS2 in the time domain between 10:00 am and 10:15 am;

(10) FIG. 4e the course of the load noise for 24 hours extracted from the course of the total load PG, PGG of FIG. 4a;

(11) FIG. 5a a forecast of the course of the basic load PS0E predicted over 24 hours without the predicted higher load oscillations PS1E, PS2E and the load noise PRE that are shown separately;

(12) FIG. 5b the superposition of the predicted course of the basic load PS0E and the predicted higher load oscillations PS1E, PS2E and the predicted load noise PRE of FIG. 5a, forming the predicted course of the total load PGE;

(13) FIG. 6a symbolically a power grid BN of a railway system EA, from which load data are transferred to a load data processor RL, which exchanges data with a system controller RBN, a planning data processor RP and an energy data processor REN;

(14) FIG. 6b the actual course of the total load PG of the power grid BN of the railway system EA up to 10:30 a.m., that is analysed to determine the current signal phases and/or amplitudes or values of the basic load PS0 according to FIG. 4b, the load components PS1, PS2 according to FIG. 4c and FIG. 4d as well as the amplitude of the load noise PR according to FIG. 4e, in order to initiate corrective measures;

(15) FIG. 7 an external energy supply system EVN2 and a control loop EAR with which an internal energy supply system EVN1 is controlled; and

(16) FIG. 8 courses of the predicted load components PS1E, PS2E of FIG. 5a, which exhibit cycle durations of 15 and 30 minutes, the average value PM15 of the sum of these load components PS1E, PS2E for a cycle duration of 15 minutes as well as correction signal kpw that is formed by subtracting the average value PM15 from the sum of the load components PS1E, PS2E and that is supplied to the control loop EAR shown in FIG. 7.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

(17) FIG. 1 shows in as an example a load curve of a total load PG recorded in a technical system EA, namely in a power grid BN of a railway system EA over 24 hours. The railway system EA is shown with two trains or system units BNE that act as loads in the power grid BN. The power grid BN of the railway system EA is supplied with electrical energy by a first energy supply system EVN1 and a second energy supply system EVN2.

(18) The diagram shows that in the time ranges around 08:00 and 18:00 the traffic volume rises and higher loads are applied to the power grid BN of the railway system EA. Around 08:00, loads of up to 500 MW and around 18:00, loads of up to 550 MW are registered. In between, the loads are reduced to a level below 400 MW. In this example it is shown that the peak loads LS1, approx. at 08:15, LS2, approx. at 18:10, and LS3, approx. at 20:00, are most significant. With peak load LS1, the load or the power consumption rises by approx. 66% from approx. 300 MW to approx. 500 MW within a short period of time. In known load management systems, loads had been deactivated for longer periods of time. Further, it is important to note that peak loads can sporadically occur also outside the typical ranges of increased power consumption. Hence, if load management systems would only consider the maximum energy requirement around 08:00 and 18:00 in the load curve, then the sporadic but intense peak load LS3 at 20:00 would be neglected.

(19) It should further be noted that the average load in the time range around 18:00 is higher than the average load around 8:00. Therefore, for handling said peak loads LS1; LS2 in these time ranges, preferably different measures are applied.

(20) According to the inventive method, for avoiding or compensating peak loads a load data processor RL (see FIG. 6a) is used that is equipped with a software program dedicated to performing the inventive load management. The load data processor RL receives, e.g. from the system controller RB of the technical system EA or the power grid BN of the railway system EA or from an energy data processor REN of the energy supply system (see FIG. 6a) data relating to the actual course of the total load PG measured in the technical system EA or in the power grid BN. Further, the load data processor RL can also process data of reported partial loads, e.g. loads of network segments, which are summed up for determining the total load PG.

(21) FIG. 2 shows that load courses that were recorded over a plurality of days (in the given example from 15 Dec. 2014 to 18 Dec. 2014) had been summed up and averaged, in order to differentiate characteristic features of the load curve from random occurrences. For a single day a single course of the total load PG is recorded. For courses of the total load PG which had been recorded for a plurality of days an averaged course of the total load PGG is calculated, which allows a more precise forecast of future load courses.

(22) From the course of the total load PG or from the averaged course of the total load PGG, individual courses of load components PS0, PS1, PS2 are determined (see FIGS. 4b, 4c and 4d), which significantly contribute to peak loads that appear in the technical system EA.

(23) A course of a basic load is preferably also determined, which corresponds to the floating average of the course of the total load PG or the averaged course of the total load PGG. The course of the basic load is determined by smoothening of time- and data series or by removing signal components with higher frequency. The load data processor RL preferably comprises a signal processor, in which a filter, preferably a FIR-filter is implemented, which serves for filtering the preferably averaged course of the total load PG, PGG.

(24) For determining periodic load components PS1, PS2 in the preferably averaged course of the total load PG, PGG preferably a Fourier transform is applied. Alternatively, band-pass filters are applied with which periodic load components PS1, PS2 can individually be detected.

(25) The left side of FIG. 2 shows the course of the total load PG, PGG in the time domain. The right side of FIG. 2 shows the course of the total load PG, PGG in the frequency domain.

(26) FIG. 3a shows load oscillations or load components resulting from the Fourier transform with the indication of the amplitude, the frequency and the cycle duration for frequencies up to 2 mHz.

(27) FIG. 3b shows load oscillations or load components resulting from the Fourier transform with the indication of the amplitude, the frequency and the cycle duration for frequencies above 2 mHz.

(28) On the abscissa of the diagram the frequency and the cycle duration of the load oscillations are entered (in FIG. 3a in minutes and in FIG. 3b in seconds). On the ordinate the amplitude of the load oscillations is indicated in megawatts. It is shown that in the range of the basic load a power consumption of several 100 MW occurs. Further, load oscillations with high amplitude and cycle durations of 1 minute, 10 minutes, 15 minutes and 30 minutes occur. For facilitating the description, in a simplified example below only load oscillations with cycle durations of 1 minute and 30 minutes are considered. These load oscillations are caused by the synchronised timetable or by the operation cycles with which the trains travel in the railway system EA.

(29) FIG. 4a shows the preferably averaged course of the total load PG, PGG over 24 hours, which corresponds for example to the course of the total load PG, PGG of FIG. 1 or FIG. 2.

(30) FIG. 4b shows the course of the basic load PS0 over 24 hours, which has been extracted from the course of the total load PG, PGG of FIG. 4a by subtracting signal components with higher frequency such as frequencies above 10 mHz. As mentioned, the course of the basic load PS0 is gained by filtering or averaging the course of the total load PG, PGG. This course is known as Moving Average.

(31) FIG. 4c shows the course of a first load component PS1 over 24 hours extracted from the course of the total load PG, PGG of FIG. 4a, which exhibits a cycle duration of 30 minutes. It can be seen that after 05:30 a rise of the amplitude of this 1.sup.st load component or load oscillation PS1 occurs that reaches a value of about 50 MW.

(32) FIG. 4d shows the envelope of a second load component PS2 over 24 hours extracted from the course of the total load PG, PGG of FIG. 4a, which exhibits the cycle duration 1 minute, and a detailed view of this load oscillation PS2 in the time domain between 10:00 and 10:15. This load component PS2 reaches amplitudes slightly above 10 MW.

(33) FIG. 4e shows the course of the load noise for 24 hours extracted from the course of the total load PG, PGG of FIG. 4a. It can be seen that the load noise can reach high power levels and is preferably also taken into account for load management purposes.

(34) After determining the different load components PS0, PS1, PS2, the future courses PS0E, PS1E, PS2E are determined for each of these load components PS0, PS1, PS2. It may be assumed that the future courses of the load oscillations remain unchanged, if the conditions do not change and the processes in the technical system EA are not altered. If no changes occur, then the predicted load oscillations PS0E, PS1E, PS2E will exactly correspond to the extracted load components PS0, PS1, PS2.

(35) FIG. 5a shows schematically the course of the basic load PS0E which has been predicted or forecast for the next day. Further shown are the courses PS1E, PS2E and PRE which have been predicted for the next day for the first two load components PS1, PS2 and the load noise PR.

(36) FIG. 5b shows schematically the superposition of the predicted course of the basic load PS0E and the predicted higher load oscillations PS1E, PS2E and the predicted load noise PRE of FIG. 5a, forming the predicted course of the total load PGE.

(37) When forecasting the course of the total load PGE, internal and/or external influences acting on the technical system EA are preferably taken into account. This can be done in several ways. On the one hand, courses of the total load PG, which have been recorded earlier under similar internal and/or external influences or conditions, can be considered when forecasting the course of the total load PGE. Alternatively, for internal and/or external influences deviations from the predicted course of the total load PGE can continuously be evaluated and taken into account when calculating or estimating the future course of the total load PGE.

(38) At lower temperatures, heating systems are normally started in the technical system EA, which leads to a high load on the energy supply system. If from one day to the other a temperature drop is forecast, then load courses can be used for predicting the course of the total load PGE, which were recorded at a day with the predicted temperatures. Alternatively, the course of the total load PGE can be evaluated and then be modified depending on the internal and/or external influences, e.g. by adding an offset.

(39) Likewise, information can be collected for occurring incidents. For each incident the impact on the load on the technical system EA is examined. Preferably, the incidents are divided into classes, to which load parameters are assigned, which are considered in the forecast of the course of the total load PGE.

(40) Hence, the method allows automatic learning for generating knowledge based on experience, which is applied for future forecasts of the course of the total load PGE or PSOE. The collected knowledge and the current and/or predicted influence parameters or values can advantageously be processed by a neural network in order to optimally predict the course of the total load PGE or PSOE.

(41) The course of the total load PGE predicted for the next day allows detecting peak loads based on reference values or thresholds th, th1, th2. Thereby, fixed thresholds th1, th2 and/or variable thresholds th can be used. Fixed thresholds th1, th2 can be applied depending on the time of the day. In the present example the lower first threshold th1 is applied between 00:00 and 12:00 and the higher second threshold th2 is applied between 12:00 and 24:00. With the first threshold th1 a first peak load LS1 is detected approx. at 08:30 and with the second threshold th2 further peak loads LS2, LS3 and LS4 are detected between 17:00 and 19:00.

(42) For the detected the peak loads LS1, . . . , LS4 control signal cs1, . . . , cs4 are generated with a predetermined lead time. With the control signals cs1, . . . , cs4 selected loads or system units of the technical system EA are deactivated or the energy data processor REN of the energy supply system EN is controlled such that the energy required for the peak loads LS1, . . . , LS4 is timely allocated.

(43) Preferably a variable or adaptive threshold th is applied, which is selected depending on a) the time of the day, and/or b) the predicted course of the total load PGE, and/or c) the energy prices EP, and/or d) the available energy reserves, and/or e) the predicted value WPGE of the total load PGE at this point in time.

(44) Most favourable is the selection of a variable threshold that is based on the predicted course of the basic load PSOE, to which an offset is added. The offset is preferably selected such that a bandwidth B results, within the predicted low noise PRE will occur.
th=WPGE+B

(45) The offset or the bandwidth B can be a product of a plurality of factors.
B=k*1/EP*1/WPGE

(46) For higher energy costs and higher values for the predicted total load PGE preferably a smaller bandwidth B is selected.

(47) By using a variable or adaptive threshold th, peak loads LS1, . . . , LS4 can easily, precisely and completely be detected.

(48) The lead time with which a crossing of the thresholds th, th1, th2 or the occurrence of peak loads LS1, . . . , LS4 is reported in advance by the load data processor RL, is selected such that the required measures, a load control action such as deactivation of system units or the allocation of energy, can be done right ahead in time. A larger lead time is preferably selected for peak loads with higher amplitude and longer duration.

(49) Corrective measures are preferably selected depending on the nature of the peak loads LS1, . . . , LS4. For peak loads with a shorter duration, measures are taken which do not interfere with the operating processes executed in the technical system EA, e.g. deactivation of heating systems. Interventions into the operating processes will take place preferably only for peak loads with an unexpected high amplitude and duration.

(50) FIG. 6a shows a power grid BN, for which load data are measured and forwarded to a load data processor RL, which exchanges data with a system controller RBN, a planning data processor RP and an energy data processor REN. The load data processor RL may receive load data from the system controller RBN or from the energy data processor REN.

(51) FIG. 6b shows the current course of the total load of the power grid BN of the railway system EA up to 10:30 a.m., that is analysed to determine the current signal phases and/or amplitudes or values of the basic load PS0 according to FIG. 4b, the load components PS1, PS2 according to FIG. 4c and FIG. 4d as well as the amplitude or power of the load noise PR according to FIG. 4e, in order to initiate corrective measures. Shifts of the clock cycle, changes of the amplitudes of individual load components PS0, PS1, PS2 and changes of the amplitude of the load noise PR, are preferably registered as early as possible, so that the predicted course of the total load PGE can be corrected and a corrected course of the total load PGEK can be established. Since peak loads may significantly change when the load components PS1, PS2 are shifted, recording of changes allows further improving the inventive load management procedures.

(52) Load control for reducing or avoiding predicted peak loads is executed centrally or peripherally. The load data processor RL communicates the amplitude and duration or the complete curve of the peak loads to the system controller RBN, whereafter the system controller RBN determines stationary or mobile system units BNE which will be controlled according to the invention. Stationary or mobile system units BNE are preferably designed such that they can be operated in a plurality of operation modes, in which different power consumption is present. In this case, the system controller RBN can forward a desired mode of operation to a peripheral controller, e.g. a train controller, whereafter the train controller switches the controlled system units BNE accordingly.

(53) The inventive method, which serves for monitoring and controlling the power grid BN of a technical system EA has been described with reference to a railway system EA. However, technical systems can comprise different devices, in which electrical loads, particularly electrical engines, are operated. The inventive method can be applied in production sites or facilities, in which high loads periodically appear or in which processes are periodically executed.

(54) FIG. 7 shows an external energy supply system EVN2 and a control loop EAR with which a local or internal energy supply system EVN1 is controlled, which supplies a first part es1 of the energy required the technical system EA. The external energy supply system EVN2, which delivers a second part es2 of the energy required by the technical system EA, is controlled by the energy data processor REN such that the basic energy requirement of the technical system EA is covered. The first energy part es1 and the second energy part es2 cover the total energy requirement of the technical system EA. The control loop EAR controls the internal energy supply system EVN1 such that the second energy part es2 delivered by the external energy supply system EVN2 does not overshoot and possibly undershoot a predetermined value. Ideally the second energy part es2 remains constant at the lowest possible level.

(55) The operator of the technical system EA and the operator of the external energy supply system EVN2 may for example close a contract for the delivery of the second energy part es2, which may correspond to the basic or average energy requirement of the technical system EA. In order to avoid deviations from the agreed maximum energy consumption from the external energy supply system EVN2, i.e. the defined second energy part es2, the internal energy supply system EVN1 is controlled accordingly to cover additional energy requirements. If deviations from the basic load or the second energy part es2 occur, which may be caused by events that are represented in FIG. 7 as disturbance variable d, then an additional energy requirement shall not be covered by costly energy delivered from the external energy supply system EVN2, i.e. by increasing the second energy part es2, but with lower-priced energy delivered from the internal energy supply system EVN1, i.e. by increasing the first energy part es1. E.g., a weather change may cause a change of the disturbance value d requiring activation of heating systems in the technical system EA. The energy required by heating systems, which would exceed the value of the second energy part es2 is then covered by increasing the first energy part es1 by controlling the internal energy system EVN1 accordingly.

(56) For controlling the technical system EA in this way, as shown in FIG. 7, a guide value fg, which corresponds to the value of the second energy part es2 and preferably to the basic load, is delivered to the control loop EAR and is compared with a measurement value mg, which corresponds to the current energy requirement of the technical system EA that is measured and/or calculated. The difference between the guide value fg and the measurement value mg is used as control deviation ra, which is forwarded to a controller RR that provides an actuating variable sg to the internal energy supply system EVN1. With the actuating variable sg the output of the internal energy supply system EVN1 is controlled in such a way that an additional energy requirement is covered by an increase of the first energy part es1.

(57) Due to the waviness of the energy requirements large load deviations may cause large deviations of the actuating variable sg in the control loop EAR, which are undesirable, since said waviness may not be fully absorbed by controlling the internal energy supply system EVN1. The remaining waviness of the load is applied to the external energy supply system EVN2 and may cause a violation of the contract and penalties.

(58) In order to further reduce the waviness, a correction signal kpw, which corresponds to the waviness of the load of the technical system EA, is evaluated and added to the guide value fg. Changes in the waviness of the load are therefore predictable and can be immediately be compensated by controlling the internal energy supply system EVN1 with a lead time. Hence, the internal energy supply system EVN1 accurately follows the waviness of the predicted load curves, e.g. the course of the total load PGE. Hence, the waviness of the load of the technical system EA is fully compensated by the internal energy system EVN1, wherefore the waviness of the load of the technical system EA has none or little influence on the second energy part es2 delivered by the external energy system EVN2. The described disturbances, which influence the technical system EA, are compensated by the control loop EAR as described.

(59) FIG. 7 shows that the correction signal kpw is subtracted from the guide value fg, so that a change of the load or energy requirement according to the waviness represented by the correction signal kpw leads to an immediate change of the actuating variable sg. The correction signal kpw can be applied with a lead time so that the internal energy system EVN1 delivers the energy without delay. In such a way, the correction signal kpw, which corresponds to the waviness, forms an anticipatory control deviation, which influences the actuating variable sg foresighted, in order to achieve a desirable control of the internal energy supply system EVN1.

(60) The correction signal kpw or waviness of the load of the technical system EA can be derived from one or more significant load components PS0, PS1, PS2. In a preferred embodiment the predicted load curve of at least one load component is taken or the predicted load curves of a plurality of load components PS1, PS2 are summed up for one or a plurality of control periods cp and the direct component or offset is removed in order to build the correction signal kpw.

(61) With the second control signal kpw, which does not comprise a DC-offset, the waviness information of the load of the technical system EA is applied to the first energy supply system EVN1, which is controlled accordingly and which is part of the control path RS the control loop EAR shown in FIG. 7.

(62) FIG. 8 shows courses of the predicted load components PS1E, PS2E of FIG. 5a, which exhibit cycle durations of 15 and 30 minutes and the average value PM15 of the sum of these load components PS1E, PS2E established for control durations or control periods cp of 15 minutes. By summing up the load oscillations PS1, PS2 and subtracting the average value PM15 for every control period cp, the above described correction signal kpw is formed, which is applied to the control loop EAR of FIG. 7 and predicts the expected control deviation that is caused by the waviness of the load with a lead time.

REFERENCES

(63) [1] J. Bosch, J. M. Aniceto, Potenziale fr das Lastmanagement im Bahnenergiesystem (potentials for the load management in a railway energy system), ebElektrische Bahnen, issue 2, 2013 [2] J. Bosch, Frequenzkomponenten des BahnstromlastgangsZusammenhnge mit dem Bahnbetrieb (load oscillations of the rail current consumptioninterdependencies with the railway operation), ebElektrische Bahnen, issue 4, 2014, [3] EP2505416A1 [4] EP2799307A1 [5] JPH0516808 [6] JPH0834268A