METHOD FOR MODELING THE OPERATION OF AN INDUSTRIAL PLANT, APPLIED TO INDUSTRIAL PLANTs, PUBLIC NETWORKS, MICRONETS AND EMBEDDED SYSTEMS

Description

GENERAL TECHNICAL FIELD AND PREVIOUS ART

[0001] The invention relates to the general field of power electronics and its applications on electrical distribution or embedded networks, smart buildings and microgrids, and more particularly to filtering and compensation modules for electrical networks and power injection from renewable sources.

[0002] It is well known that the quality of the electrical signal delivered by the distribution network has a direct consequence on the performance of the systems powered by this network, as well as the lifespan of the electrical equipment that constitutes the network or is connected to it.

[0003] In particular, harmonic distortion, phase shift between voltage and current, i.e. the reactive, as well as the imbalance of currents and voltages are factors that allow us to express the quality of the energy passing through this network.

[0004] Imbalance is understood to mean a difference in the physical quantities of the signal between the different phases and / or amplitudes, for example voltage levels, intensity.

[0005] Total harmonic distortion is a measure of the linearity of signal processing performed by comparing the output signal of a device to a sinusoidal input signal.

[0006] The non-linearity of the system distorts this sinusoid. The output signal remains a periodic signal that can be analyzed as a sum of sinusoids with frequencies that are multiples of the frequency defining the period, called the fundamental frequency.

[0007] Each of these sinusoids is a harmonic whose rank is equal to the quotient of its frequency divided by the fundamental frequency. The total harmonic distortion is the ratio of the RMS values ​​of the fundamental frequency to the other frequencies.

[0008] A device with nonlinear loads connected to an electrical network receives power from the electrical network and re-injects a signal into the network, the signal re-injected into the network being degraded by the operation of the device.

[0009] A non-linear load is defined as a load based on power electronics components that draws active power (with or without reactive power) from the power supply system and, more importantly, feeds distorting power (related to harmonics) back into the power grid. These harmonics can be of conventional order (5, 7, 11, 13, etc.) for 3-wire power systems: three-phase non-linear loads (widely used in industrial areas). In the case of single-phase non-linear loads installed in a 4-wire power system (3 phases and neutral), commonly used in residential, commercial, or administrative areas, 3rd-order harmonics and their odd multiples (3, 9, 15, etc.) will circulate in this network in addition to the conventional harmonics.

[0010] On the other hand, the phase shift between the voltage and the signal intensity implies the appearance of reactive power which leads, among other things, to a decrease in the active power that can be transmitted in the network.

[0011] The development of energy production technologies, particularly from renewable sources, has led to the emergence of numerous individual production units, especially in so-called positive energy housing solutions.

[0012] When a power production unit, particularly in the case of renewable energies such as solar or wind, is connected to the network, it is necessary to use one or more inverters or rectifier-inverters in order to manage and convert the power generated by the unit in direct current and direct voltage into an alternating signal, before injecting it into the network.

[0013] Inverters, most often consisting of bidirectional, open- and closed-controlled power electronic switches, such as IGBTs (IGBTs, from the English Insulated Gate Bipolar Transistor or insulated-gate bipolar transistor) and GTOs (GTO, from English Gate Turn-off Thyristor or trigger-extinguishing thyristor). By a set of appropriately controlled switches (usually pulse-width modulation), the source is modulated to obtain an alternating signal of the desired frequency.

[0014] The switching frequencies of the power electronics components of inverters generally induce high-frequency harmonic components in the signal injected into the network, which degrades the quality of the signal passing through the network.

[0015] In a set such as that represented in figure 1a, comprising a general network 1, a non-linear and / or linear load 2, and a continuous energy storage element 3 connected to an inverter 4, it is known to connect at the output of the inverter 4 a filter 5 configured to block the high-frequency switching components contained in the signal to be injected into the electrical network 1.

[0016] Thus, the high-frequency harmonics of the signal caused by the switching of inverter 4 are not transmitted via filter 5 to network 1; only the desired signal is generated by inverter 4 and injected into network 1 or load 2.

[0017] Different filter structures 5 have been proposed for this purpose.

[0018] A first-order filter, the most commonly used; composed of a simple inductor with practically negligible internal resistance, as shown in figure 1b , in particular, does not allow this function to be fulfilled.

[0019] Indeed, the higher the inductance of such a filter, the greater its ability to prevent switching components from being injected into network 1. However, the higher the inductance, the slower the rate of change of the current flowing through the filter, and the greater the phase shift the filter will cause between the desired actual current and the current injected into the network 1 - load 2 system via filter 5.

[0020] Conversely, a low value of inductance allows the majority of components due to switching to be injected into network 1 and consequently to affect electrical installations and equipment.

[0021] The correct sizing of the first-order output filter will therefore depend on the compromise to be found between the dynamics and the efficiency of the inverter-based device, especially when it operates as a parallel active filter with harmonic pollution control.

[0022] This compromise is very difficult to achieve without the use of an auxiliary passive filter 6 installed at the output of the output filter 5 or upstream on the network side, to filter the high-frequency components as shown in figure 1b .

[0023] However, this auxiliary filter 6 can cause unwanted side effects such as resonance with other passive elements installed on the electrical network 1.

[0024] These electrical resonance phenomena sometimes result in voltage or current peaks far exceeding the permissible values ​​for devices connected to network 1, causing the destruction of these devices.

[0025] This auxiliary filter also causes active power consumption due to its damping resistor. Furthermore, the filtering quality of these auxiliary filters degrades over time due to the aging of their passive components. GENERAL PRESENTATION OF THE INVENTION

[0026] One aim of the invention is to "clean up" the current consumed by a non-linear load by canceling, on the grid side, its harmonic content, unbalanced content, and reactive content of the current, with the objective of improving voltage quality on the grid side. This invention, having a 4-wire structure, is adapted to the harmonic spectrum containing the conventional harmonics (5, 7, 11, 13, etc.) of industrial areas, as well as those of order 3 and their odd multiples (3, 9, 15, etc.) of residential, commercial, or administrative areas.

[0027] Another goal is to maximize the active power output of a renewable energy generation unit. Figures 1a, 1b .

[0028] Another goal is to optimize the energy consumption of a smart building.

[0029] Another goal is to optimize the energy production of a conventional energy production unit (from fossil fuels: oil, gas, etc.) within a microgrid.

[0030] Another goal is to optimize the management of energy flowing between production units (renewable and conventional) and consumption units.

[0031] To achieve this, the invention proposes a method for modeling the electrical operation of an industrial site which includes a plurality of machines controlled and distributed over one or more production lines according to claim 1.

[0032] Optionally but advantageously, in such a process, the machines with their controls, based, among other things, on power electronic converters, are modeled by current sources, representing the total consumption (active, reactive and harmonics, etc.) of the machines, measured by the power quality analyzer installed at the entrance of each production line.

[0033] More specifically, the modeling steps are as follows: 1- Install a measuring device which is a spectrum analyzer / power quality analyzer, at the input of each production line of the site studied, and which provides all the electrical quantities of the network, namely: active, reactive and apparent power, power factor, fundamental components of current and voltage, total harmonic distortion (THD) of current and voltage, individual harmonic distortion of currents and voltages, etc.2- Model the active and reactive power by a parallel RL branch or by a single current source that represents the fundamental current, with an angle corresponding to the power factor; the direction of current flow is from the electrical network 1 to the load 2 (represented here by the current sources), 3- Model the individual harmonic distortion rate of the currents and voltages by current sources; each corresponding to a harmonic order, the direction of harmonic current flow is from the load (2) (represented here by the current sources) to the electrical network (1), 4- Connect all these current sources to one or more resistors of very high values ​​(on the order of MΩ), to force the currents to flow in only one direction. PRESENTATION OF THE FIGURES

[0034] Other features and advantages of the invention will become apparent from the following description, which is purely illustrative and not limiting, and should be read in conjunction with the accompanying figures in which: there figure 1a is a diagram representing an electrical network supplying a load and a renewable energy generation power unit connected to an inverter, via one or more capacitive storage elements, including a filter according to the prior art; the figure 1b represents this same network with a different filter structure from the prior art; the figure 2 is a diagram representing an electrical network supplying a load with an inverter, included in an active compensation device, as an example; the figure 3 is a diagram representing the structure of an active compensation device according to an example, specifically the structure of the control unit; the figure 4is a diagram representing an electrical system according to an example, in which the compensation device is inserted, in parallel, between a renewable energy production unit, within a self-consumption building, and the entire grid-load system; the figure 5 is a diagram representing an electrical system according to an embodiment of an example, in which the system is connected to a smart building; the figure 6 represents the regulation loop of the capacitive storage element with the objective, among others, of extracting the maximum active power from the renewable energy generation unit; the figure 7 represents the integration of the calculation method for this maximum active power into the algorithm for identifying / calculating inactive currents (harmonics, reactive and unbalanced, etc.); the figure 8is a diagram representing an electrical system according to a second embodiment of the invention, in which the electrical system comprises a semi-centralized management system that manages the conventional power generation of the grid and, in case of emergencies, the operating mode of the centralized units via active compensation devices connected, in turn, to the renewable power generation units; the figure 9 represents a general structure of the parallel active filter according to an example; the Figure 10 represents a phase-equivalent diagram of the third-order LCL output filter according to an example; the figure 11 represents the gain diagram of the third-order LCL output filter; the figure 12 represents a general block diagram of the active filter current control algorithm; the figure 13 represents an effect of phase shift on the compensation quality of the active filter; more specifically the figure 13arepresents the intensity of the charge, the figure 13b represents the setpoint current and the injected current, and the figure 13c represents the network current with the effect of the phase shift as well as the estimated ideal current without the effect of the phase shift; the figure 14 represents the gain and phase diagrams of the transfer function of the control loop of the active filter with RST, which is a linear controller based, among other things, on the placement of the poles of the closed loop (control loop); the figure 15 represents the pursuit and single-phase control signal with sliding-mode controllers using the Sign, Sigmoid, and artificial relative degree augmentation control functions; the figure 16 represents the pursuit and single-phase control signal with higher-order sliding mode controllers using the SMC, C-HOSM, 2-SMC Twisting, 2-SMC Super-Twisting, and Lyapunov approach control algorithms; the figure 17represents an electrical diagram of the industrial site under study; the figure 18 represents a photograph of the power quality analyzer installed at the industrial site under study; the figure 19 represents a model of the overall load of the production line of the studied factory; the figures 20a to 20i represent the measurement curves delivered by the power quality analyzer; more specifically the figure 20a represents the three-phase traces (Ph1, Ph2 and Ph3) of the total load currents IL123 in A; the figure 20b represents the voltages of the electrical network Vs123 in V; the figure 20c represents the power factors PF123; the figure 20d represents the total harmonic distortion (THD) of the total load currents THDi123 in %; the figure 20e represents the total harmonic distortions of the THDv123 network voltages in %; the figures 20f to 20irepresent respectively the individual harmonic distortion rates for the 5th, 7th, 11th and 13th harmonic distortions of the load currents for each phase, with the distortion rates expressed as a percentage; figures 21a to 21f represent the simulation results for validating the proposed model; in particular the figure 21a represents the total load currents on phase ph1, including harmonics; the figure 21b represents the total load currents on phase ph2, including harmonics; the figure 21c represents the total load currents on phase ph3, including harmonics; the figure 21d represents the simple network tensions; the figure 20e represents the power factors; the figure 21f represents the harmonic distortion rates on the different phases, including THD-V and THD-I; the figure 22represents the filtering of the current in the studied disturbed network using sliding mode controllers with Sign, Sigmoid, and artificial relative degree augmentation control functions; the figure 23 represents the filtering of the voltage of the network under study using sliding mode controllers with Sign, sigmoid and artificial relative degree augmentation control functions; the figure 24 represents the filtering of the current and voltage of the disturbed network under study using the C-HOSM controller; the figure 25 represents the filtering of the current and voltage of the disturbed network under study using the 2-SMC Twisting controller; the figure 26 represents the filtering of the current and voltage of the studied disturbed network using the Super-Twisting controller; the figure 27 represents the current-voltage and power-voltage characteristics of a photovoltaic generator; the figure 28represents a DC voltage regulation loop for the inverter storage element unit; the figure 29 represents an algorithm for identifying disturbing (non-active) currents incorporating the pursuit of the maximum power point. DESCRIPTION OF ONE OR MORE IMPLEMENTATION AND REALIZATION METHODS

[0035] An example concerns an electrical system as represented in figures 2-3 , comprising a 3-phase supply network 1 and a neutral, a non-linear or linear load 2 or both connected to network 1, a capacitive energy storage element 3, a renewable energy generation power unit 100 and a compensation set 7 connected on one hand at its input, downstream, to the output of the renewable energy generation power unit 100 associated with the storage element 3 and on the other hand in parallel to its output, upstream to a connection point (C) located between network 1 and load 2.

[0036] Compensation set 7 includes: a power conversion unit 8, comprising an inverter 9, generating an alternating current with a frequency band from 50 to 2500Hz, covering: a) the entire frequency band of the non-active disturbing current which exhibits: all or part of the harmonics, and at the fundamental frequency all or part of the reactive power and / or the unbalanced current, etc.b) as well as the active current generated by the renewable energy generation power unit 100; a filtering unit 10, comprising at least one LCL type filter 11, combining two inductors and a capacitor, for each of the phases and a neutral, and connected: on the one hand downstream of the voltage inverter 9, and on the other hand in parallel to the connection point C between the given electrical network 1 and the non-linear and linear electrical loads 2, the LCL output filter 11 being sized to block the harmonic components due to the switching of the inverter 9; a control unit 12 configured to identify and control the current injected into the network 1 and to control the inverter 9 to generate this current.

[0037] The control unit 12 is configured to control the inverter 9 so that the inverter 9 generates a signal of intensity I inj configured in opposite phase to cancel, at the connection point C, the non-active disturbances of the signal generated by the load 2 and injected into the network.

[0038] The signal emitted at the output of the filtering unit 10, and therefore at the output of the compensation assembly 7, therefore presents the harmonics of the signal caused by the non-linear part of the load 2.

[0039] For any harmonic present in the disturbing signal, caused by load 2, the signal emitted by the filtering unit 10 has a voltage value inverse (in opposite phase) to the voltage value of the signal from load 2 non-linear at that harmonic n.

[0040] Thus, by injecting the output signal of the compensation assembly 7 at the connection point C, the harmonics of the signal emitted by the non-linear load 2 are canceled and the current flowing on the network side 1 is free of harmonics.

[0041] Similarly, the output signal of the compensation assembly 7 has, for the fundamental frequency and the harmonics, a phase shift configured to generate a reactive power inverse (in opposite phase) to the reactive power of the disturbing signal of the load 2.

[0042] In this way, the reactive power consumed by load 2 is completely compensated, within the limits of the sizing of device 7, on the network side.

[0043] The control unit 12 is further configured to compensate for the imbalance of the signal passing through the non-linear or linear load 2.

[0044] Imbalance is understood to mean a different current consumption depending on the different phases of several single-phase loads.

[0045] It should be noted that any interfering current at a frequency included in the frequency band of the injected current control loop, ranging from 0 Hz to 2500 Hz, can be compensated / filtered by device 7 ( cf. Fig. 14 ). Unconventional harmonics may be included.

[0046] The compensation assembly 7 is configured to be installed in an electrical network already polluted at the level of its upstream voltage while offering the possibility of operating at two fundamental frequencies (50 and 60 Hz), which can each vary by ± 10%, while being able to adapt to voltage variations of ± 15%.

[0047] This allows the compensation system 7 to function correctly even in the case of generator sets replacing the electrical grid.

[0048] The control unit 12 is further configured to regulate (charge and maintain constant) the voltage V dc across the terminals of the capacitive storage element 3.

[0049] The control unit 12 is further configured to detect the maximum power point (MPPT) Maximum Power Point Tracking ) of a renewable energy generation power unit 100 and generate, via the inverter 9, the active current corresponding to this maximum power; the voltage V dc across the terminals of the capacitive storage element 3 is equal, in this case, to the voltage of the maximum power of the renewable energy production unit 100. The conversion unit 8

[0050] The conversion unit 8 comprises, as represented in figures 2 And 3, the 9-phase, 3-phase, neutral inverter has a structure similar to a conventional two-level voltage inverter, in order to ensure reliability and minimize the costs of the structure.

[0051] Inverter 9 is here controlled by control unit 12 so as to occupy, with units 3 and 11, a parallel active filter function.

[0052] The 4-wire, 4-arm inverter 9 has, for each of the 4 arms, two switching devices 13 and 14, each connected on one side to an output terminal of the capacitive storage element 3 and on the other side to a common connection point A to which one of the 4 wires is also connected.

[0053] This connection point located, for each arm, between the first switching device 13 and the second switching device 14 forms a connection point of one of the 4 wires of the inverter 9 connected to the other end of the filtering unit 10 whose output is connected to the connection point C.

[0054] Each switching device consists of a power electronics switch D, controllable to open and close (IGBT, GTO or others) and an antiparallel diode B (bidirectional operation).

[0055] The inverter 9 emits, on each wire, a signal with an intensity I inj.

[0056] Inverter 9 is configured to inject power that covers a wide frequency band, usually from the fundamental frequency, for example 50 Hz, up to the 50th harmonic, which in the example of a 50 Hz fundamental has a frequency of 2500 Hz. The filtering unit 10

[0057] The filtering unit 10 includes in this embodiment an LCL type filter 11, which has the advantage of being easy to implement compared to a first-order filter, which is bulky, heavy and expensive.

[0058] The filter 11 therefore comprises, on each of the 4 wires (three phases with the neutral), a first inductance 15 connected on one side to an input wire of the filtering unit 10 (coming out of one of the 4 common connection points A), and on the other side to a second inductance 16, the second inductance 16 being connected on the other side to one of the 4 connection points C via an output wire of the filtering unit 10.

[0059] The filter 11 associated with each wire also includes a capacitor 17 connected on one side to a common point E and on the other side to a connection point F between the first inductance 15 and the second inductance 16.

[0060] The filter 11 prevents the propagation of components due to the switching frequency of the inverter 9 to the supply network 1 without degrading the dynamics of the compensation system 7. It therefore limits the risk of resonance, and thus eliminates the need for an auxiliary passive filter intended to limit this propagation, in the case of using a first-order output filter (a simple inductance) associated with this auxiliary filter. The control unit commands 12

[0061] The control unit 12 is configured to regulate the voltage across the storage element 3 and to identify and control the injected current, via the filtering unit 10, at the connection points C for the 4 wires, and to control the inverter 9 to generate this current. The identification of the currents to be injected at the connection points C is performed by the reference current calculation unit 25.

[0062] The current to be injected at the connection points C includes the active current of the renewable energy generation power unit 100 as well as disturbing non-active currents, which may have harmonics, imbalance, and reactivity, etc. configured in opposite phase to oppose the harmonics, imbalance and reactivity of the signal passing through the load 2 in order to reduce them, or even cancel them out on the electrical network side 1.

[0063] Controller 23 ensures the continuation of the current between the injected current I inj and the reference current from unit 25. For its part, controller 62 ensures the continuation of the voltage between the voltage V dc across the terminals of the storage element 3 and the maximum power voltage of unit 100, delivered by unit 25-B; unit 3 will therefore be charged.

[0064] Finally, the control unit 21 ensures the control of the inverter switches 9 in order to generate I inj.

[0065] The operation and structure of the control unit 12 will subsequently be described for one phase. It is understood that each of the 4 wires (the three phases as well as the neutral) is controlled in a similar and independent manner by the control unit 12, which therefore comprises a control chain for each wire.

[0066] For the control of the switching devices 13, 14 of one wire, the control unit 12 therefore has two outputs, one connected to the triggering pole of the first switching device 13, the other being connected to the triggering pole of the second switching unit 14.

[0067] The two outputs are connected together to the output of a logic comparator 18, so as to simultaneously control the switching of the two switching devices 13, 14.

[0068] One of the outputs of the control unit 12 has a logic level inverter 19, so that the activation of one of the switching devices 13, 14 causes the deactivation of the other, advantageously during a single control pulse thanks to their common connection, upstream of the logic level inverter 19, to the same comparator 18.

[0069] The comparator 18 emits a logic output signal corresponding to the comparison between an output signal from a saturation element 20 and an output signal from a switching control device 21.

[0070] The switching control device 21 controls the switching of the inverter 9 via a non-linear controller 23 by continuous sliding mode, adapted to the LCL type filter 11 which ensures in closed loop control of the entire frequency band from 50 to 2500 Hz for the injection of the non-active disturbance current and the active current, according to the identification of the reference currents by the calculation unit 25.

[0071] The switching control of inverter 9 is implemented so as to allow the following to pass through the output filtering unit 10 at the connection point C: a part or all of the non-active disturbance currents injected in opposite phase comprising: harmonic currents, as well as reactive and unbalanced currents at the fundamental frequency, in the non-linear and linear electrical loads 2, to satisfy the non-active energy consumption demand of the non-linear and linear electrical loads 2, while decontaminating the electrical network 1 of these non-active disturbance currents; an active current corresponding to a maximum available power point within the renewable energy generation power unit 100, to satisfy the active energy consumption demand of the non-linear and linear electrical loads 2.

[0072] The switching control device 21 emits a signal (called a carrier wave) at a predetermined frequency, depending on the sizing of units 9, 10 and 3, so as to cause the switching of the switching devices 13, 14.

[0073] In most cases, the control laws, designed to control voltage inverters that are connected to the electrical grid via an LCL filter, were originally established for renewable energy systems (photovoltaic and wind).

[0074] In this case, linear controllers can be used because the inverter only has fundamental components (primarily active power) to inject into the grid. The phase shift, at the fundamental frequency, between the signal identified as Iref-fundamental via the reference current calculation unit 25-B and the injected signal Iinj-fundamental is negligible in this case. It should be noted that in this operating mode, unit 25-A, due to a phase shift issue, is not included in the identification block, and unit 25 is thus limited to unit 25-B.

[0075] Indeed, in the case of a parallel active filter, the inverter 9 must inject both fundamental (reactive and unbalanced) and harmonic components into the electrical grid, which implies complete control over a very wide frequency bandwidth. A significant phase shift limits the applicability of linear controllers in the case of harmonic filtering.

[0076] Non-linear control of the injected current I inj makes it possible to overcome this phase shift problem, encountered when using an LCL filter associated with a linear controller.

[0077] In this embodiment, the controller 23 performs a sliding mode control (in English Sliding mode control or SMC), which ensures a desired dynamic response, strong robustness / insensitivity to bounded disturbances, and good control properties over a wide range of operating conditions.

[0078] It is well known that a classic sliding mode control with a so-called Sign function ( sign (in English), generates a very high frequency control of the inverter switches (discontinuous control), in order to ensure finite-time convergence to the slip surface where the system states are subsequently maintained, even in the presence of bounded disturbances.

[0079] The sign function of a signal is given by the relation: sign u = u u

[0080] In practice, the voltage inverter is controlled by a frequency-limited switching function, either fixed or variable. Switching at very high frequencies causes overheating that can ultimately destroy the inverter.

[0081] To avoid discontinuous control operation, this invention employs two continuously controlled sliding mode controllers, one associated with a sigmoid-approximate sign function and the other with a method for artificially increasing the relative degree, followed by an integrator. The two controllers are associated with two suitable sliding surfaces.

[0082] The first method consists of replacing the Sign function with a continuous approximation. This involves approximating the Sign function with a Sigmoid function given by the formula: sign u = u u + ε with ε the thickness of the area surrounding the sliding surface.

[0083] In this case, the system no longer converges towards the desired value, but towards a neighborhood of it, which makes it possible to limit switching at very high frequencies.

[0084] The method for artificially increasing the relative degree consists of increasing the order of the sliding surface by one degree (leading to an additional derivative). This allows an integrator to be inserted at the output of the Sign function, and the control in this case becomes continuous. Indeed, by differentiating once again the tracking error between the signal Iref, identified by unit 25, and the injected current Iinj, the signal at the output of the Sign function represents the derivative of the control (which was discontinuous until now). An integral action at the output of the Sign function will generate a continuous control.

[0085] The two controllers used make it possible to avoid a discontinuity in the control, in particular caused by variable switching of the inverter at very high frequencies, caused by a classic sliding mode control.

[0086] In addition, three other nonlinear controllers based on the higher-order sliding mode method are also employed; these are the second-order sliding mode control algorithm 2-SMC Twisting and 2-SMC super-Twisting as well as the continuous higher-order sliding mode control algorithm C-HOSMC.

[0087] These controllers ensure, in a wide range of operating conditions, as described below, the robustness in performance and stability provided by the classic discontinuous sliding mode control method, while offering continuous control which ensures nominal operation of the inverter (9).

[0088] Optionally but advantageously, the switching control device 21 includes a pulse width modulation device, in which the control is compared to a carrier wave with a predetermined switching frequency depending on the sizing of units 9, 10 and 3.

[0089] This allows the inverter 9 to operate at a fixed switching frequency.

[0090] A synergistic effect is obtained by thereby improving the retention of high-frequency components, due to switching, by the LCL type filter 11.

[0091] A saturation element 20 is conventionally configured to impose upper and lower limits on the control signal. These limits are determined by the amplitude of the carrier wave, which is in turn related to the voltage of the capacitive storage element 3.

[0092] The input signal of the saturation element 20 comes from the output of a summing 22, whose output signal is the sum of a control voltage u issued by a controller 23 and the voltage Vs of the connection point C. The addition of the voltage Vs in the control loop of the injected current Iinj prevents a strong reactive current inrush from the inductors of the LCL filter 11.

[0093] The reference current calculation unit 25 comprises a plurality of measured inputs.

[0094] The measured quantities of the system include: the current intensity IL of each wire consumed by load 2, the current intensity I inj of each wire exiting the filter unit 10, the voltage V dc across the terminals of the capacitive storage element 3, which is also the maximum output power voltage of the renewable energy generation power unit 100, the voltage V s of each phase at the connection points C, via a phase-locked loop (PLL) 26. The phase-locked loop 26 is used to extract the three-phase positive component V d123 of the grid voltage 1 at the connection points C. It should be noted that the three-phase positive component V d123 of the grid voltage (1) is essential for the efficient operation of the identification unit 25. the current intensity I pv of the renewable energy generation power unit 100, the voltage V pv of the renewable energy generation power unit 100% renewable

[0095] The reference current calculation unit 25 has an output for each phase, each output corresponding to the control chain of the associated phase (including the neutral), said control chain comprising a comparator 24, a controller 23, a summing 22, a saturation element 20 and a logic comparator 18 whose output is divided into two branches, one of which includes a logic level inverter 19.

[0096] The reference current calculation unit 25 therefore emits a setpoint signal per phase, including the neutral, presenting an intensity I ref.

[0097] The intensity I inj of each phase, including the neutral, of the signal emitted by the inverter 9 is returned to the control unit 12 and compared, via the comparator 24 with the setpoint I ref of the control chain of the corresponding phase, delivered by the unit 25. The difference between I ref and I inj is corrected via the controller 23 which issues the command u.

[0098] The voltage V dc applied across the terminals of the capacitive storage unit 3 is regulated (maintained constant) by comparing it, via comparator 60, with a reference voltage V dc-ref, which is equal to the maximum power point voltage V MPP of the renewable energy generation power unit 100, calculated via unit 25-B.

[0099] The measured voltage signal V dc is filtered from fluctuations at 300 Hz or other, via a second-order low-pass filter 61. The error signal (V dc-ref - V dc) is controlled by a PI (Proportional Integral) controller 62 or other, in order to obtain the maximum power P MPP.

[0100] The voltage VS of network 1 is added in the summing 22 to the control u, the output of the summing 22 being limited in the saturation element 20, the output of the saturation element 20 being compared via the logic comparator 18 with the signal delivered by the pulse width modulation device 21 (the carrier).

[0101] The output signal of comparator 18 is at level 1 if the output signal of the saturation element 20 is greater than the carrier. Otherwise, it is at level 0. Following this logic chain, the switching device 13 or 14, which does not include a logic level inverter 19, is respectively closed or open (the other device operating in a complementary manner).

[0102] The control unit 12 includes at least one processor and at least one memory, the memory comprising a program executed by the processor so as to implement the process of determining the setpoint signal I ref, containing the non-active current as well as the current I MPP of the maximum power point MPPT, via the reference current calculation unit 25, of the injected current control I inj via the control unit 23, of regulating the voltage V dc across the terminals of the storage unit 3 via the controller 62 and of generating this injected current by controlling the devices 13, 14 of each wire of the inverter 9 via the control unit 21.

[0103] The process for determining the setpoint signal I ref, delivered by the reference current calculation unit 25, comprises the following steps: processing of measurements (based on the algorithm for identifying instantaneous or other 25-A power) of the current IL of the load 2 and the voltage at the connection point Vs in order to estimate the non-active harmonic, reactive and unbalanced currents, etc.; generation of a setpoint I ref calculated to cancel (in part or in full depending on the sizing of unit 7) the non-active content of the current Is on the supply network side 1.

[0104] Optionally, but advantageously, the reference current calculation unit 25 is configured to identify, via unit 25-B, the maximum power point (MPPT) of operation of the renewable energy generation power unit 100, installed within a renewable energy production field, building, or self-consumption plant 99, as follows: processing of current Ipv and voltage Vpv measurements (based, via unit 25-B, on a P&O type algorithm from English) "Disturb and Observe" or others, which is part of unit 25) of the renewable energy generation power unit 100. generation of the maximum voltage V MPP, via unit 25-B, and consequently the maximum power P MPP and the maximum current I MPP which will be integrated into the setpoint current I ref via the power P MPP calculated at the output of controller 62.

[0105] Advantageously, only the maximum power point calculation algorithm (unit 25-B) needs to be integrated into unit 25, via the disturbance current identification algorithm (unit 25-A). This is because the inverter capacitor voltage regulation loop (unit 3) is already included in unit 25-A to charge the inverter capacitor during operation as a pure active filter, thereby compensating for losses caused by the inverter components and the LCL filter. In this invention, and during operation as an active filter with maximum power injection from a renewable energy source, the reference voltage Vdc-ref becomes the maximum power voltage delivered by unit 25-B, instead of being predefined according to specifications oriented towards pure filtering.This methodology lightens, minimizes and improves accuracy, in a significant way, during the implementation of the control part, compared to the state of the art which provides an additional current loop which also requires the use of a chopper.

[0106] The control unit 12 is further configured to regulate the voltage V dc across the terminals of the capacitive storage element 3, and consequently ensure the charging of the capacitor 3. Indeed, the capacitor 3 has, among other things, the role of covering the losses of units 9 and 10 as well as providing the maximum active current I MPP of the renewable energy generation power unit 100 to the connection points C, via the inverter 9 and the filtering unit 10.

[0107] The next step in the process within Unit 12: processing of the measurements of the voltage V dc across the terminals of the capacitive storage element 3; filtering, via a second-order low-pass filter 61, of the measured signal of the voltage V dc of fluctuations at 300 Hz or others; tracking, via the controller 62, between V dc-ref =V MPP and the voltage V dc, in order to ensure a constant maximum voltage V MPP across the terminals of the capacitive storage element 3; generation of the disturbance current injected in opposite phase to that flowing in the load and of the maximum power current of the unit 100.

[0108] For this pursuit via controller 62, the reference current calculation unit 25 is configured to integrate the calculation of the maximum power current, via unit 25-B, into the algorithm for calculating non-active currents, provided by unit 25-A.

[0109] The integration process is as follows: the output of controller 62 being the maximum power P MPP of unit 100, this signal is added to the summing 63, which has the active disturbing power at its second input. P̃ from unit 25-A (Upstream) of the calculation of instantaneous disturbing powers (active P̃ responsive Q̃ and homopolar P 0). The 25-A (Upstream) unit calculates, from the voltages Vs at the connection points C and the currents IL of the load 2, the instantaneous disturbing powers (active P̃ responsive Q̃ and homopolar P 0 in the α, β and 0 frame) caused by non-active disturbance currents present in the current of load 2 IL.

[0110] The calculation of the setpoint / reference currents (Iref) is performed via a reverse calculation, from the 25-A unit (upstream) through the unit (25-A downstream). This reference current includes the non-active currents as well as the maximum power current, first calculated in the same α, β, and 0 coordinate system and then in the three-phase, 4-wire system.

[0111] Units 25-A(Upstream) and 25-A(Downstream) are configured to identify disturbing currents; units 25-A(Upstream), 25-A(Downstream) and 25-B represent the calculation unit for reference currents 25.

[0112] Next, unit 12 having the setpoint current I ref, from unit 25, as well as the injected current I inj controlled in turn via controller 23, inverter 9 is controlled via control unit 21 to generate the current I inj.

[0113] This avoids the need for an additional device, a chopper, to generate the maximum power from the renewable energy generation unit 100.

[0114] In this case, the inverter imposes the voltage on the DC side. This represents an additional reliability for the parallel active filter device as well as a significant cost saving.

[0115] Optionally, the control unit 12 may include a chopper configured to maintain a predetermined DC voltage across the inverter's energy storage element (3). This voltage may correspond to the voltage of a battery bank to be charged. For example, the control unit 12 may include: A chopper configured to maintain a predefined constant DC voltage across the inverter's energy storage element 3, regardless of the voltage level of the renewable energy source 100, to ensure unchanged harmonic filtering. A predefined fixed voltage can charge a battery bank, serve as a common DC bus for a microgrid, etc. A dual chopper is used for islanded power grids. The second chopper can provide, for islanded grids and other applications, an additional DC voltage level, including for charging a battery bank. In an islanded grid, the inverter sets the voltage and frequency on the AC side, while the second chopper sets the voltage on the DC side.

[0116] In one embodiment, the compensation device 7 is installed within a so-called smart building 27, that is to say that the electrical devices contained in the smart building 27 can, among other things, be selectively controlled and activated by a decentralized management unit 70, for example to operate during so-called off-peak periods of the day, during which the energy demand of the network is low and the cost of energy, from the point of view of the consumer, decreases.

[0117] The control unit 12 is connected to the decentralized management unit 70 of the smart building 27. The decentralized management unit 70 communicates in real time to the control unit 12 the power levels of the loads (electrical devices) in the operating or non-operating state of the building 27.

[0118] The control unit 12, having in real time the maximum power of the renewable energy generation power unit 100 delivered by the unit 25-B and the controller 62 as well as the powers of the loads in operation or not of the building 27 communicated by the decentralized management unit 70, is configured to regulate the consumption of the different devices according to at least two modes of economic load distribution.

[0119] In a first mode of economic load distribution, the smart building 27 comprising a renewable energy generation power unit 100 and in the event of a high availability of primary renewable sources, the control unit 12 is configured to control the decentralized management unit 70 in such a way as to activate most of the devices of the smart building 27 according to the production of the unit 100; this consumption mode is called the adapted mode.

[0120] In a second mode where renewable primary energy is scarce, the control unit 12 is configured to control the decentralized management unit 70 in such a way as to selectively activate the devices of the smart building 27, so as to present a flat load curve or others, depending on the power producible from the network; this consumption mode is called modulated mode.

[0121] This helps to avoid peak demand on network 1 and to benefit from a reduced cost of energy consumed.

[0122] In this case, most or even all of the energy consumed is from renewable sources, which minimizes energy consumption from network 1. In the event of a high demand for supply and in addition to the second distribution method and if the electrical network 1 is intelligent, a third distribution method adapted to the production of the intelligent electrical network 1 is conceivable.

[0123] In one embodiment, the compensation device 7 is installed between network 1 and an industrial site or a residential, administrative or commercial building, which can be considered from the point of view of network 1 as a non-linear / linear disturbing load 2.

[0124] Before installing such a device on an existing industrial site, it is sometimes necessary to demonstrate the qualities of the device to the site owner, who will then decide whether or not to equip their site with the device.

[0125] It is therefore necessary to model the site in order to simulate its performance with and without the device, in order to appreciate the contribution of the compensation device.

[0126] Traditionally, industrial site modeling is carried out by producing a complete model containing all the machines and their controls (based on power electronic converters) present in the area studied.

[0127] However, this type of modeling faces two problems: hardware and computer science, which limit its adoption as a credible model for validating solutions applied to this type of electrical network.

[0128] Indeed, in this case, as many measuring devices are required as there are controlled machines. Furthermore, it proves difficult to simulate a multi-level industrial network containing several power transformers, numerous electrical lines and cables, and a plurality of controlled machines with multiple pollution control filters.

[0129] In this embodiment, a single measuring device is installed at the entrance of each production line.

[0130] The site modeling is carried out by replacing the controlled machines with their currents (active, reactive and harmonics) measured by one of the three-phase measuring devices.

[0131] The simulation model is greatly simplified because current sources, the number of which is limited by a maximum harmonic rank theoretically set at 50, replace the model of controlled machines.

[0132] It should be noted that a current source is always associated with a very high resistance (on the order of MΩ) connected in parallel with the current source.

[0133] The model is then validated by comparing the simulation results with measurements taken at the input of the production lines.

[0134] The model is finally used to credibly simulate the effect of the compensation device 7 on the signal passing through the loads of the industrial site and on the disturbing current flowing in the network 1.

[0135] In one embodiment, the system comprises a plurality of renewable energy generation power units 100, for example of the wind or photovoltaic type, each connected, directly or via a device 7, to a local network 28, itself connected to network 1.

[0136] Local network 28 corresponds to the actors producers, consumers and consumer-producers (of buildings with self-consumption or positive energy).

[0137] In most cases, each production unit 100 is connected to the local network 28 by means of a compensation device 7 configured to maximize the power output of the production unit 100 and to prevent the propagation of electrical disturbances, if present, from the upstream side of the compensation device 7 to the local network 28 and the electrical network 1.

[0138] Each compensation device 7 is connected, electronically, to a semi-decentralised management system 29, to which it communicates, in real time, information concerning the consumption as well as the actual and forecast energy production of the production units 100.

[0139] In addition, the semi-decentralised management system 29 receives real-time information on the energy output of the conventional (fossil) energy production units 80, installed within the local network 28.

[0140] The semi-decentralised management system 29 manages the economic allocation of conventional (fossil) energy generators within the local network 28, based on the total renewable energy production and the total consumption within the same local network 28.

[0141] The semi-decentralised management system 29 receives in real time information on renewable and conventional power available at any time on the local network 28.

[0142] The system also includes a plurality of consumption points, for example a smart building or a smart industrial site, each consumption point being assimilated from the point of view of the network to a non-linear / linear load 2, each of the consumption points being connected to the local network 28, is equipped with a compensation device 7 configured to depollute the current flowing on the local network side 28 of disturbances caused by the load 2 and to control the activation of the different devices of the consumption point via the decentralized management unit 70.

[0143] Each compensation device 7 is connected to the semi-decentralised management system 29, to which it communicates information concerning instantaneous consumption and future consumption according to the programmed operation of the consumption points, in order to estimate the energy demand.

[0144] The semi-decentralised management system 29, receiving at all times all the data (actual and forecast) of the energy that can be produced as well as that to be consumed by the different actors of the local network (producer, consumers, consumer-producers), can intervene, only when necessary when the total demand for energy within the local network 28 is much greater than the total production, with the pilots of the decentralised control units of the smart buildings 27 to switch the consumption mode to modulated mode (flat load curve) in favour of the global system.

[0145] The term "consumer-producer station / actor" refers to a self-consumption or net-positive energy building. The semi-decentralized management system 29 is therefore configured to estimate, over a given period, the total energy demand of the local network 28 that it monitors.

[0146] Based on power production and energy demand estimates, the semi-decentralized management system 29 is configured to: To perform, in real time, the economic allocation of production to the conventional energy production units 80, to intervene, only when necessary when the total energy demand within the local network 28 is significantly greater than the total production, with the decentralized control units 70, via the control units 12 of the devices 7, to switch the consumption mode to modulated mode (flat load curve). To control, as a last resort and in case of risk of a production shortfall, the circuit breaker 90 to allow energy consumption from network 1. This operating mode is not possible when the local network is islanded, i.e., not connected to network 1.

[0147] The management system 29 is therefore empowered to prioritize the operation of certain consumption points over others, in order to distribute energy demand over the given period.

[0148] Furthermore, the semi-decentralized management system 29 can distribute power demand over time so that, when power from network 1 is needed, it is consumed during periods of low demand in order to minimize costs and avoid loading network 1 during peak demand.

[0149] It should be noted that the maximization of the power of the renewable energy production units 100 as well as the optimization of consumption, within a smart building 27, are ensured locally by the devices 7 via the control units 12. The semi-decentralized management system 29 is called so because it is only called upon to ensure the economic distribution of production of the conventional energy production units 80 and to correct consumption when necessary.

[0150] Following a study concerning the invention carried out by the inventors, the inventors wish to make the following comments. Comparison of Invention and Prior Art 1. General structure of the prior art parallel active filter General Topology

[0151] There Fig. 9This presents the general structure of the parallel active filter, which consists of two blocks: the power circuit and the control circuit. The power circuit comprises: of a voltage inverter based on power switches, controllable at turn-on and turn-off (GTO, IGBT, ...etc), with diodes in antiparallel, of a capacitive energy storage circuit, of an output filter.

[0152] The control circuit, for its part, consists of: of the method for identifying disturbed currents, of the PLL-based system (phase-locked loop or in English "Phase Locked Loop" (in English) which is integrated into the method of identifying disturbing currents, regulating the DC voltage applied to energy storage elements, controlling the current injected into the network from the voltage inverter, and controlling the voltage inverter. Output filter

[0153] The output filter is a passive filter used to connect the voltage inverter to the electrical grid. The output filter is sized to meet the following two criteria: to ensure the dynamics of the current d dt I → h − L = d dt I → inj with I h_L the harmonic current contained in the load current IL and I inj the current of the active filter injected into the network, prevent the components due to switching from propagating on the electrical network.

[0154] Two types of output filter can be used: a first-order output filter and a third-order output filter. First-order filter

[0155] This type of filter is the most used in the literature; it consists of a simple inductance Lf with practically negligible internal resistance.

[0156] A filter of this type cannot simultaneously satisfy both output filter sizing criteria. Indeed, only a relatively low value of Lf can achieve good active filter dynamics while satisfying the above equation.

[0157] Unfortunately, a low value of Lf allows the majority of switching components to end up on the network side and consequently affect electrical installations and equipment.

[0158] Conversely, a relatively high value of L f will prevent these components from propagating on the electrical network, but will affect the dynamics of the active filter and thus degrade the quality of compensation.

[0159] The correct sizing of the first-order output filter will therefore depend on the compromise to be found between the dynamics and the efficiency of the parallel active filter.

[0160] This compromise is very difficult to achieve without the use of an auxiliary passive filter installed at the output of the inverter or upstream on the grid side.

[0161] However, this auxiliary filter can cause unwanted side effects such as resonance with other passive elements installed on the electrical network.

[0162] It also causes active power consumption through its damping resistance.

[0163] Furthermore, the filtering quality of these auxiliary filters degrades over time due to the aging of their passive elements. Third-order filter (LCL): modeling in the s-plane

[0164] The third-order output filter is an alternative to a heavy, bulky, and expensive first-order filter, allowing us to escape the problems mentioned in the case of the first-order output filter.

[0165] This output filter consists of two inductors (Lf1, Lf2) with respective internal resistances (Rf1, Rf2) and a capacitor Cf with a small damping resistance Rf (see the Fig. 10 ) which we will disregard later.

[0166] It should be noted that (L s , R s and es ) represent respectively the inductance, resistance and electromotive force of the upstream electrical network.

[0167] This type of filter, thanks to the additional degree of freedom provided by the capacitance C f, can ensure the two output filter sizing criteria that we mentioned earlier.

[0168] The equations that model the output filter are: I inj = B 1 s A s V f s + B 2 s A s V s s with V f the inverter output voltage, B 1 (s) / A(s) the transfer function of the output filter with the network corresponding to the original system (to be controlled) and B 2 (s) / A(s) the transfer function corresponding to the disturbance model.

[0169] These disturbances are caused by the voltage of the electrical network e(s) which is now considered equal to the connection voltage Vs for high power short-circuit electrical networks. A s = a 1 s 3 + a 2 s 2 + a 3 s + a 4 B 1 s = b 11 s + b 12 B 2 s = − b 21 s 2 + b 22 s + b 23 with a 1 = L s + L f 2 L f 1 C f a 2 = L s + L f 2 R f 1 C f + R s + R f 2 L f 1 C f + L f 1 + L s + L f 2 R f C f a 3 = L s + L f 2 + L f 1 + R s + R f 2 R f 1 C f + R f 1 + R s + R f 2 R f C f a 4 = R f 1 + R s + R f 2 b 11 = R f C f b 12 = 1 b 21 = L f 1 C f b 22 = R f + R f 1 C f b 23 = 1

[0170] From the previous relationships, if we neglect all resistances (except Rf), we can establish the following relationship, valid at frequencies above 50 Hz: B 1 s A s = R f C f s + 1 L f 1 L f 2 C f s 3 + R f C f L f 1 + L f 2 s 2 + L f 1 + L f 2 s

[0171] The resonance frequency fcp of the LCL filter, if we neglect the resistance Rf in this case as well, is given by the relation: f cp = 1 2 π L f 1 L f 2 C f L f 1 + L f 2

[0172] The LCL output filter is sized to reject components due to the inverter switching frequency, which has been set at 16 kHz, to match an industrial application case that we propose.

[0173] Thus, a rejection of more than -50 dB is obtained for a cutoff frequency of 1900 Hz.

[0174] This choice allowed us to effectively attenuate the high-frequency components, as shown in the output filter gain vs. frequency diagram. Fig. 11 . Third-order filter (LCL): modeling in state space

[0175] Based on the Laplace representation (Maths. 1), the representation of the LCL filter in the state space can be given by the following equations: dv c dt = i f − i inj C f di f dt = 1 L f 1 v f − v c − R f i f − i inj − R f 1 i f di inj dt = 1 L f 2 v c + R f i f − i inj − v s − R f 2 i inj

[0176] The vector-matrix representation is given by: x ˙ = A x + B v f + P v s with A = − R f 1 + R f L f 1 R f L f 1 − 1 L f 1 R f L f 2 − R f 2 + R f L f 2 1 L f 2 1 C f − 1 C f 0 , B = 1 L f 1 0 0 , P = 0 − 1 L f 2 0 x = i f i inj v c 2. Linear control by RST of the state-of-the-art voltage inverter connected to the grid via the LCL filter RST Controller

[0177] The control strategy is based on estimating current disturbances using an identification algorithm. Then, the voltage inverter, controlled by PWM (Pulse Width Modulation): Pulse width modulation "), generates the currents injected into the network I inj, which must follow the identified reference currents I ref ( I inj → I ref ). The closed control loop is designed to ensure high-precision tracking.

[0178] The general diagram of the current control system is shown on the Fig. 12In this diagram, the voltage inverter (controlled by PWM) is connected to the power grid via an LCL filter, with an RST-type controller, and the instantaneous power method or other methods are used for identifying current disturbances. A phase-locked loop (PLL) is used to extract the positive component of the grid voltage (Vd) to achieve the expected performance of the identification algorithm employed.

[0179] The mains voltage Vs represents an external disturbance, the effects of which are compensated by adding the same mains voltage to the control signal (u). This will prevent the reactive fundamental current from flowing from the mains to the active filter via the inductances of the LCL filter.

[0180] From equation (1) and using an RST controller, we obtain: I inj s = T B 1 s S A + R B 1 s I ref s + S B 2 s S A + R B 1 s V s s

[0181] With R ( s ), T ( s ) And S ( s) are the polynomials of the controller.

[0182] The order of R ( s ) And S ( s ) is the same as the system order LCL B 1 s A s ; therefore, polynomials R ( s ) And S ( s ) are of third order. T ( s ) is chosen such that I ref s I inj s = 1 for the entire frequency band (50-2500 Hz) included in the reference signal I ref ; T ( s ) can be a simple gain in this case.

[0183] It should be noted R s S s represents the transfer function of the RST controller; the common denominator D ( s ) = ( ITS + RB 1)( sThe arbitrary stability polynomial (ASP) contains the closed-loop poles. These poles are placed in a sector of (2 × 45°) to ensure a damping of 0.7. Finally, the control loop poles are positioned to ensure a fast and accurate response with good disturbance rejection. It should be noted that the pole values ​​are limited by the closed-loop cutoff frequency. Effect of phase shift

[0184] The RST controller, as well as all linear controllers, can be used when the references to be tracked consist of constant signals or signals at a single, relatively low frequency (as in the case of reactive power compensation or imbalance at the fundamental frequency of 50 Hz). At this frequency, the phase shift between the identified references ( I ref ) and the output of the injected closed loop ( I inj) is acceptable. However, if the reference signal to be tracked is composed of signals at several frequencies, the phase shift is no longer negligible. Indeed, the phase shift increases with frequency. The effect of the phase shift of the structure shown in Fig. 12 is presented on the Fig. 13 From this figure, we can observe that the disturbed current ( I Load ) is not well compensated for ( I really : compensation with phase shift), relative to the ideal form ( I ideal : phase-shift free compensation).

[0185] The control loop transfer function with the RST I inj ( s ) / I ref ( s ) is presented via the Fig. 14 .

[0186] The gain and phase of the closed-loop transfer function I inj ( s ) / I ref ( s ) are given via Table I, for multiples of the fundamental frequency ranging from 50 Hz up to (23×50) Hz.It should be noted that we limited ourselves to 1150 Hz because electrical networks self-filter, via their inductance, high-order harmonic currents.

[0187] It should be noted that the RST linear controller provides unity gain (0 dB ) for virtually the entire frequency band (50-2500 Hz: cf. Fig. 4 Furthermore, the RST is only used to compensate for unbalanced and / or reactive currents at the fundamental frequency.

[0188] Indeed, at this frequency, a phase shift of (-1°) is negligible. Above this frequency, the phase shift is no longer negligible, and the parallel active filter cannot compensate for the harmonic currents. 3. NON-LINEAR CONTROL (FIRST-ORDER SLIDING MODE) OF THE VOLTAGE INVERTER ACCORDING TO THE INVENTION CONNECTED TO THE NETWORK VIA THE LCL FILTER

[0189] In most cases, the control laws designed to control voltage inverters connected to the electrical grid via an LCL filter were originally developed for renewable energy systems (photovoltaic and wind). In this case, linear controllers can be used because the inverter only has fundamental components (active power) to feed into the grid: the phase shift is negligible in this scenario.

[0190] However, in the case of a parallel active filter, the inverter must inject both fundamental (reactive and unbalanced) and harmonic components into the electrical grid, which implies complete control over a very wide frequency bandwidth. It should be noted that a significant phase shift limits the applicability of linear controllers in the case of harmonic filtering.

[0191] The alternative to overcome this phase shift problem, encountered when using an LCL filter with a linear controller, is nonlinear control. In this invention, we employ the sliding mode control (SMC) method. "Sliding Mode Control" (in English) to ensure a desired dynamic response, strong robustness / insensitivity to bounded disturbances, and good control properties over a wide range of operating conditions.

[0192] On the other hand, it is well known that a classic SMC controller generates a command, inverter switches, at very high frequency, in order to ensure a finite-time convergence to the slip surface where the system states are maintained for all the following time, even in the presence of bounded disturbances.

[0193] In practice, the voltage inverter is controlled by a limited / fixed frequency switching function. Therefore, variable switching at very high frequencies, if it occurs, could lead to overheating that could ultimately destroy the inverter.

[0194] Therefore, we use a classic SMC (with function sign ) à Continuous control. Indeed, in order to avoid discontinuous control (variable inverter switching at very high frequencies), caused by a classic SMC, we opt for two sliding mode controllers, one of which is associated with a function sign approximate depending on sigmoid and the other is with a method we call AIRD (an artificial increase in relative degree, followed by an integrator, in English: “an Artificial Increase Relative Degree, followed by an integrator”). Finally, the active filter control generated by the continuous SMCs will be modulated by a PWM, in order to allow the active filter to operate at a fixed switching frequency, suitable, on the one hand, for nominal operation of the inverter's power electronics components and easy, on the other hand, to filter by the LCL, which facilitates, among other things, the blocking of high-frequency components by the LCL filter. Design of a classic sliding controller

[0195] Consider system (3) with: y := i inj representing the entrance, u := vf is the control signal and w := vs as disturbances, and for R f = 0, equation (3) becomes: x ˙ = Ax + Bu + Pw , y = Cx , C = 0 1 0 with A = − R f 1 L f 1 0 − 1 L f 1 0 − R f 2 L f 2 1 L f 2 1 C f − 1 C f 0 , B = 1 L f 1 0 0 , P = 0 − 1 L f 2 0

[0196] It is clear that the relative degree of system (4) is r = 3, because CB = CAB = 0 and CA 2 B = 1 L f 1 L f 2 C f , then a sliding variable is chosen in the form: S = K 0 e + K 1 e ˙ + K 2 e ¨ withe = i inj - i ref . It should be noted ė, ë are obtained using a higher-order sliding mode differentiator.

[0197] The positive coefficients K 0, K 1 and K 2 of equation 5 are selected to make the system (Maths. 4) exponentially stable with the desired convergence rate, in the sliding mode defined by S = 0 .

[0198] In order to design an SMC controller that derives S → 0 in finite time and which preserves the states of the system (Maths. 4) in a sliding surface S = 0 For the entire period that follows, the dynamics of the sliding variable are derived: S ˙ = F di ref dt d 2 i ref dt 2 v s d v s dt d 2 v s dt 2 i inj i f v c + K 2 L f 1 L f 2 C f u with F di ref dt d 2 i ref dt 2 v s dv s dt d 2 v s dt 2 i inj i f v c = K 0 e ˙ + K 1 e ¨ − K 2 d 3 i ref dt 3 + − R f 2 L f 2 2 − 1 L f 2 C f R f 2 L f 2 + R f 2 L f 2 2 C f i inj − R f 2 L f 2 2 C f + R f 1 L f 1 L f 2 C f i f + R f 2 L f 2 2 1 L f 2 − 1 L f 2 2 C f − 1 L f 1 L f 2 C f v c − R f 2 L f 2 2 − 1 L f 2 C f 1 L f 2 v s + R f 2 L f 2 dv s dt − 1 L f 2 d 2 v s dt 2

[0199] It is assumed that: the disturbance V s with its first two derivatives d v s dt , d 2 v s dt 2 as well as d 3 i ref dt 3 are bounded, the first two derivatives of the pursuit error e ,ë as well as the variables I inj , if , vc are bounded within a reasonable state domain, which includes the operating point.

[0200] So, there is η > 0 such that F di ref dt d 2 i ref dt 2 d 3 i ref dt 3 v s dv s dt d 2 v s dt 2 i f i inj v c ≤ η ,

[0201] The condition for the existence of a sliding mode S . S ˙ < 0 can be easily achieved by the SMC u : = v f = − λ sign S with u the command applied, via PWM, to the inverter, λ is fixed and defined by the saturation limiter block at 420V (see Fig. 12 Next, the action of the control on S ˙ is proportional to K 2 L f 1 L f 2 C f u which explains the choice of K 0, K 1 and K 2 in (7): K 0 = 306000000, K 1 = 35000 and K 2 = 1

[0202] In this context, it is important to point out that the Sign function in (Maths. 8) has been approximated in terms of a Sigmoid function sign S = S S + ε with ε = 2 × 10⁸ SMC with artificially increased relative degree (AIRD)

[0203] We extend the system (Maths. 4) as follows: x 4 = u , x ˙ 4 = v with v the new control, the system, in this case, becomes x ˙ = A e x + B e v + P e w , y = C e x , C e = 0 1 0 0 And A e = − R f 1 L f 1 0 − 1 L f 1 1 L f 1 0 − R f 2 L f 2 1 L f 2 0 1 C f − 1 C f 0 0 0 0 0 0 , B e = 0 0 0 1 , P e = 0 − 1 L f 2 0 0

[0204] Given that the relative degree r = 4 of the system (9), the sliding variable is chosen of the form: S e = K 0 e + K 1 e ˙ + K 2 e ¨ + K 3 e ⃛ with e = i inj - i ref .

[0205] The time derivative of the sliding surface (Maths. 10) is obtained: S ˙ e = K 0 e ˙ + K 1 e ¨ + K 2 e ⃛ + K 3 F i f i inj v c d 4 i ref dt 4 + K 3 1 L f 1 L f 2 C f v with F = − d 4 i ref dt 4 − R f 1 L f 1 L f 2 C f + R f 2 L f 2 2 C f di f dt + − R f 2 L f 2 2 − 1 L f 2 C f R f 2 L f 2 + R f 2 L f 2 2 C f di inj dt − R f 2 L f 2 2 − 1 L f 2 C f 1 L f 2 − 1 L f 1 L f 2 C f dv c dt − R f 2 L f 2 2 − 1 L f 2 C f 1 L f 2 dv s dt + R f 2 L f 2 2 d 2 v s dt 2 − 1 L f 2 d 3 v s dt 3

[0206] It should be noted e , ë And (in Maths 11) are obtained using a higher-order sliding mode differentiator.

[0207] The positive coefficients K 0, K 1, K 2 and K3 are selected, in this case as well, to ensure exponential dynamic stability of the system (Maths. 9) with the desired convergence rate in the sliding mode defined by S e = 0 .

[0208] In this case, we assume that all disturbances and their derivatives, all derivatives of the tracking error, and all state variables of the system are bounded in a reasonable state domain, which includes the operating point.

[0209] So, there is η 1 > 0 such that S ˙ e v = 0 ≤ η 1 and the condition for the existence of a sliding mode S . S ˙ < 0 For S e ≠ 0 can be easily achieved by the SMC v : = − λ sign S e where the value of λ = 10 7< is chosen to limit the effect of perturbations caused by the additional derivative resulting from the artificial increase in the relative degree.

[0210] Finally, a continuous order u = ∫ v dtis generated and limited by the 420V saturation block (see Fig. 12 This is reflected in the scheme of the Fig. 3 by an integrator at the output of the SMC controller. It should be noted that S ˙ e is proportional to K 3 L f 1 L f 2 C f v which explains the choice of K 0, K 1, K 2 and K 3 in (Maths 11): K 0 = 1.7576 e + 13 , K 1 = 2.0280 e + 09 et K 2 = 78000 et K 3 = 1 4. NON-LINEAR CONTROL (HIGHER-ORDER SLIDING MODE) OF THE VOLTAGE INVERTER ACCORDING TO THE INVENTION CONNECTED TO THE NETWORK VIA THE LCL FILTER

[0211] In this present invention, the use of higher-order continuous sliding mode control algorithms followed by pulse-width modulation (PWM) is proposed, guaranteeing a predefined switching frequency for controlling the components of the voltage inverter associated with the LCL filter. for the first time.

[0212] It should be noted that the document [D2: ALALI MHD AE ET AL: "A Lyapunov approach based higher order sliding mode controller for grid connected shunt active compensators with an LCL filter", 2017 19TH EUROPEAN CONFERENCE ON POWER ELECTRONICS AND APPLICATIONS (EPE'17 ECCE EUROPE)] proposes a higher-order Lyapunov control in which controller continuity is not guaranteed: the control is discontinuous (see Fig. 16 ), as described on page 7 of this publication.

[0213] Moreover, this command is impossible to implement, firstly because of its discontinuity, and also because of its dependence on disturbances caused, among other things, by the voltage of the network which is usually disturbed, especially in industrial areas.

[0214] It should be noted that Lyapunov's control approach is suited to systems that process constant signals (at zero frequency) in a stable environment (free from disturbances). Indeed, addressing the problem of harmonics (with a frequency band including 50-2500 Hz) in a disturbed environment (electrical networks in general, and especially industrial areas) while ensuring the stability of the active filter results in discontinuous control.

[0215] In this present invention and for the first time, Three continuous sliding mode control algorithms of second and higher order are employed. These are the second-order sliding mode control algorithms 2-SMC Twisting and 2-SMC super-Twisting, as well as the continuous higher-order sliding mode control algorithm C-HOSM.

[0216] These controllers provide, across a wide range of operating conditions, the desired dynamics of the sliding system, high insensitivity to corresponding bounded disturbances, and improved stabilization accuracy. It is worth noting that the C-HOSM and 2-SMC Twisting and Super-Twisting algorithms provide finite-time convergence to the sliding surface.

[0217] Therefore, the control unit 12 can thus include at least one of the following three higher-order sliding mode controllers: The first is based on a continuous higher-order sliding mode controller C-HOSMC, with a 1st-order adapted sliding surface, associated with three Sign functions, two of which act directly on the control and the third acting on the part of the control related to disturbances; the second employs a 2-SMC Twisting controller with artificial increase of the relative degree, with a 2nd-order adapted sliding surface, followed by an integrator, associated with two Sign functions, acting on the derivative of the control respectively via the sliding surface and its derivative; the third is a 2-SMC Super-Twisting controller with a 2nd-order adapted sliding surface, without artificial increase of the relative degree, associated with a Sign function, acting directly and by its integral term on the control. C-HOSM Controller Design

[0218] The C-HOSM control algorithm can control systems with an arbitrary relative degree. Here, the sliding variable is selected as follows:

[0219] The relative degree is r = 2 according to the equation (Maths.4). The origin of the equation (Maths.14) is as follows. In a noisy measurement environment, the sliding variable will converge towards a domain whose size is proportional to the noise amplitude w ( t ) of the measurement Therefore, the actual sliding mode S = e ˙ + c e = w t and the effect of noise on tracking error e is attenuated by means of a low-pass filter with a cutoff frequency equal to c (in our case, c = 10 4< ).

[0220] The dynamics of the sliding variable The input-output relationship is calculated according to the equations. (Maths.4), (Maths.14) S ¨ = v + f x t

[0221] The algorithm's control law C - HOSM is designed as follows: v = v 1 − v 2 Or : v 1 = − γ 1 S α 1 − γ 2 S ˙ α 2 Or S α 1 = S α 1 sign S

[0222] With : γ 1, γ 2 > 0 are selected so that the polynomial p 2< + γ 2 p + γ 1. either Hurwitz, with a desired pole placement, ( α 1, α 2) are calculated such that α 1 = α 2 − α , α 2 = α , α ∈ (0,1); if α = 1 / 2 therefore, α 1 = 1 3 , α 2 = 1 / 2 ; v 2 is chosen for a disturbance rejection such that: v 2 = − ω Or:

[0223] It should be noted that the control law in (Maths. 16)-(Maths. 19): tender , S ˙ towards zero in finite time, is continuous and therefore, is a second-order continuous sliding mode controller ( C - HOSM), knowing that the continuity of this command is ensured without artificially increasing the relative degree.

[0224] Calculation of γ 1 , γ 2 : They are calculated as coefficients of a second-order polynomial; the eigenvalues ​​of this polynomial are chosen to provide a given transient response while being limited by the switching frequency of the voltage inverter, imposed by the PWM carrier.

[0225] Calculation of λ , β They are calculated as follows: λ = 1.5 L , β = 1.1 L

[0226] In our case, L is equal to 3 × 1018

[0227] Finally, the u command applied to the inverter components is given: u = L s + L f 2 L f 1 C f v with ( L s + L f 2) L f 1 C f = a 1 of the transfer function B 1 ( s ) / A ( s ) of (Maths. 1). 2-SMC Twisting Controller Design

[0228] The sliding variable is chosen such that: S ˙ = γ 1 e + γ 2 e ˙ + e ¨ , γ 1 , γ 2 > 0

[0229] Then, S ˙ = v + φ where φ( x, t ) represents the cumulative disturbances of the system presented by (Maths. 4). By differentiating the derivative of the sliding variable S ˙ , we obtain: S ¨ = v ˙ + φ ˙ with φ̇ assumed bounded in a reasonable state domain of the system presented by (Maths. 4).

[0230] The 2-SMC Twisting control system, which makes it tender , S ˙ to zero in finite time, is designed according to the derivative of the control v : v ˙ = − α β sign S + sign S ˙

[0231] Calculation of γ 1 , γ 2 : They are selected as coefficients of a Hurwitz polynomial; the eigenvalues ​​of this polynomial are chosen according to this criterion, while being limited by the switching frequency of the voltage inverter, imposed by the PWM carrier.

[0232] Calculation of α, β : with α, β > 0; They are respectively equal to 10 8< , 0.6. They are chosen to ensure a fast response as well as an acceptable reduction of the very high switching frequency.

[0233] Finally, the control signal applied to the voltage inverter components is u = ∫ 0 t v ˙ dt

[0234] It should be noted that the command u in (24) is continuous, whereas v̇ features a very high frequency (discontinuous) control. 2-SMC Super-Twisting Controller Design

[0235] The sliding variable is chosen such that: S = γ 1 e + γ 2 e ˙ + e ¨ , γ 1 , γ 2 > 0

[0236] Then : S ˙ = u + f x t with f ( x, t ) represents the sum of the disturbances of the system, presented in (Maths. 4), whose derivative is assumed to be bounded.

[0237] The 2-SMC Super-Twisting control system, which makes it tender , S ˙ zeroing in finite time is designed such that: u = − λ S 1 / 2 − v with v ˙ = − β Sign S and v = − β ∫ 0 t sign S dt

[0238] Calculation of γ 1 , γ 2 : As in the two previous cases, these parameters are calculated as coefficients of a second-order Hurwitz polynomial; therefore, the eigenvalues ​​of this polynomial are chosen to provide a fast response, while being limited by the switching frequency of the voltage inverter, imposed by the PWM carrier.

[0239] Calculation of λ , β : They are chosen as follows: λ = 1.5 L , β = 1.1 L

[0240] With a value of L sufficiently high ḟ ( x, t) ≤ L, L being in our case equal to 1 × 10⁷ < . It should be noted that the Super-Twisting control law (Maths.27)-(Maths.29) generates a continuous control. SIMULATION RESULTS OF THE CONTROLLED INVERTER ACCORDING TO THE INVENTION Simulation results with Matlab-Simulink First-order continuous SMC controllers

[0241] The simulations are initially performed using a simple Simulink diagram. In this diagram, the identified reference harmonic currents I ref are modeled by current sources of ranks 5, 7, 11, 13, 17, and 19, which represent the same harmonic spectrum of the current that we will analyze through a case study below. The simulations are performed first using classical SMC, then SMC with a sigmoid function, and finally with SMC associated with the AIRD algorithm.

[0242] There Fig. 15 trace the single-phase currents I ref, I inj(identified and injected respectively) as well as the inverter control signal u. We observe that very precise tracking is ensured by the discontinuous control (at a variable and very high switching frequency) of the classical SMC as well as by the continuous control (at a switching frequency fixed by the PWM) of the AIRD algorithm. However, the SMC method with a sigmoid function provides slightly degraded but largely acceptable tracking. Higher-order SMCs

[0243] The simulations are carried out, for the same characteristics of the previous case, using this time, first the classical SMC and then, in order, the C-HOSM, the 2-SMC Twisting, the 2-SMC Super-Twisting and finally the Lyapunov approach of document D2.

[0244] There Fig. 16 trace the single-phase currents I ref, I inj(identified and injected respectively) as well as the control signal of the inverter u. We can see that very precise tracking is ensured by the five controllers, with a discontinuous control signal for the classical SMC and for the Lyapunov approach, while the three proposed higher-order SMC controllers ensure continuous control. Simulation results with Matlab®, Simulink®, and Simscape Power System® (Case Study)

[0245] This is the Melaiece textile factory located in the Sheikh Najjar industrial zone in Aleppo, Syria; the factory's electrical network is shown via the Fig. 17 .

[0246] The production line at this factory contains 56 asynchronous motors, 40 of which are controlled by power inverters and whose rated power varies between 0.25 and 75 kW.The plant's power factor is 0.76 without reactive power compensation, then it rises to 0.98 with compensation via controlled capacitor bank banks. LOW VOLTAGE NETWORK MODELING

[0247] Given the large number of machines ordered in an industrial site (40 in our case), it becomes difficult, if not impossible, to model the electrical networks of these installations.

[0248] The first constraint is logistical; indeed, as many measuring devices are required as there are machines being controlled (40 in our case: these are the power quality analysis devices, represented in figure 18 ).

[0249] Furthermore, it proves difficult to perform a simulation of a multi-level industrial voltage network containing several power transformers with long cables and power lines, as well as dozens of machines controlled and associated with several active pollution control filters.

[0250] The idea here is to replace machines controlled by their currents (active, reactive and harmonics) delivered by analysis devices.

[0251] In this case, an analysis device will be installed at the entrance of each production line, therefore only one measuring device per production line instead of one measuring device per machine.

[0252] From a computer science point of view, the simulation scheme is no longer cumbersome because current sources, the number of which is limited by a maximum harmonic rank, theoretically set at 50, will replace the model of dozens of controlled machines.

[0253] It should be noted that a current source is always associated with a very high resistance (of a few M Ω) connected in parallel.

[0254] There Fig. 19 shows the sources of the three-phase currents, which model the overall energy consumption of the production line of the factory studied: the fundamental current of phase Ph1 is 578 A, with a power factor of FP= 0.76, lagging.

[0255] The direction of current flow is adopted according to the convention of the direction of power flow in an electrical network: harmonic currents are disturbing, therefore they are injected by the so-called polluting load, while the fundamental current is supplied by the network.

[0256] It should be noted that we can replace the fundamental currents with a three-phase RL load; the sources of the fundamental currents I1 will, in this case, be removed from the previous model. Validation of the proposed model

[0257] In order to model the production line of the textile factory under study, we perform electrical measurements at the input of the production line, as shown in the Fig. 18 .

[0258] THE Figures 20 present these measures.

[0259] There figure 20a represents the three-phase traces (Ph1, Ph2 and Ph3) of the currents of the total load IL123 in A.

[0260] There figure 20b represents the voltages of the electrical network Vs123 in V.

[0261] There figure 20c represents the power factors PF123.

[0262] There figure 20d represents the total harmonic distortions THD of the total load currents THDi123 in %.

[0263] There figure 20e represents the total harmonic distortions of the THDv123 network voltages in %.

[0264] THE figures 20f to 20irepresent respectively the individual harmonic distortion rates for ranks 5, 7, 11 and 13 of the load currents for each of the phases, the distortion rates being expressed as %.

[0265] It should be noted that these measurements were limited by the maximum harmonic rank 50 (50× 50 Hz).

[0266] Finally, the parameters of the electrical network, presented in Fig. 17 , are quantified in Table II. Table II. Technical specifications of the electrical components constituting the network processed Electrical Network es, Ssc, Rsc, Lsc 20kV, 500MVA, 0.253Ω, Ls=0.002425H Power Transformer S n , ΔP Cu , ΔPo, u cc ,I o 1MVA, 10.5kW, 1.7kW, 6%, 1.3% Electrical Cable 3×300 mm², Len=20 m Xcable=0.16Ω / km, Rcable=0.059 Ω / km

[0267] Based on the measures presented in figures 20a to 20i We choose the moment when the voltage harmonic distortion is highest; this is the moment (22h:05:00). The state of the electrical network at this moment is summarized in Table III. Table III. State of the electrical network at the chosen time DATE Time Hz 03 / 10 / 2009 22:05:00 49.98 Ph 1 Ph 2 Ph 3 V s (V) 224.5 223.8 224.2 HE HAS) 522.19 518.3 507.1 P Load (W) 114379,2 114054,2 110071,88 Q Load (VAR) 18812,18 15860,68 24532,81 S Load (VA) 116769,1 115968,4 113611,6 PF 0,97 0,98 0,96 THD-V s (%) 6,7 6,2 6,8 THD-I L (%) 11,6 10,7 12 Ih5 (%) 11 10,4 11,5 Ih7 (%) 2,8 1,7 2,4 Ih11 (%) 1,7 1,2 1,4 Ih13 (%) 1,4 0,9 1,1 Ih17 (%) 1 0.5 1.1 Ih19 (%) 0.9 0 0.7

[0268] In Table III, the quantities P Load, Q Load, S Load, respectively are the active, reactive and apparent powers of the overall load of the production line.

[0269] Based on actual measurements and the technical specifications presented respectively via the Fig. 20 and Table II, simulations, in the time and frequency domains, were carried out.

[0270] Figures 21a to 21g represent the following three-phase values: there figure 21a represents the currents and harmonics of the total load on phase ph1 (IL1 = 520.7841 A), The figure 21b represents the currents and harmonics of the total load on phase ph2 (IL2 = 518.38 A), the figure 21crepresents the currents and harmonics of the total load on phase ph3 (IL3 = 507.13 A), the figure 21d represents the simple voltages of the network (Vs1 = 223.9 V, Vs2 = 223.07 V, Vs3 = 223.68 V), the figure 21e represents the power factors (PF1=0.967, PF2= 0.978, PF3= 0.959), the figure 21f represents the harmonic distortion rates on the different phases, and in particular the THD-V (THD-Vs1= 6.46%, THD-Vs2= 5.9%, THD-Vs3= 6.65%), as well as the THD-I (THD-I1= 11.65%, THD-I2= 10.76%, THD-I3= 11.95%).

[0271] THE figures 21a to 21c allow us to estimate the individual distortion rate of the harmonic currents of the total load: I5-ph1,2,3= (11.02%, 10.42%, 11.51%), I7-ph1,2,3= (2.72%, 1.78%, 2.36%), I11-ph1,2,3= (1.71%, 1.27%, 1.36%), I13-ph1,2,3= (1.5%, 1.08%, 1.17%), I17-ph1,2,3= (1.13%, 0.61%, 1.17%), I19-ph1,2,3= (0.94%, 0%, 0.72%).

[0272] By comparing the simulation results with the measurements presented in the Fig. 20and Table III, it is observed that all quantities (actual and measured) are almost identical, with a maximum error of 3%.

[0273] These results validate the proposed models, and more specifically the model for nonlinear loads (machines controlled by power converters, such as inverters, rectifiers, etc.). Based on these results, we can consider the following results to be valid as well. Results of simulations of the active filter installed in the network of the studied factory First-class SMC controllers

[0274] In this part, we will validate the use of the active filter associated with the LCL filter with the control algorithms, the classical SMC, the SMC with a sigmoid function as well as the SMC with the AIRD method.

[0275] The technical specifications of the components constituting the active filter are presented via Table IV. TABLE IV. TECHNICAL DATA SHEET OF THE PARALLEL ACTIVE FILTER Parallel active filter L f1 = 90 µH, R f1 = 5 mΩ LCL output filter L f2 = 100 µH, R f2 = 5 mΩ C f = 130 µF, Storage capacitor C = 0.6 mF, Vdc = 840 V LCL cutoff frequency 2000 Hz PWM switching frequency 16 kHz

[0276] There Fig. 22 presents the simulation (before and after filtering) of three-phase currents on the network side I s 123, as well as the THD of the phase 1 current ( THD - I s ).

[0277] The active filter is primed after 5 periods of the sector (up to 0.1 s); it will operate, first for 3 periods (up to 0.16 s) with the classic SMC (discontinuously controlled).

[0278] After that, for 3 periods (up to 0.22 s), it will continue with the continuously controlled SMC (associated with a sigmoid function).

[0279] Finally, after 0.22 s and until the end of the simulation (0.28 s), the active filter will be controlled by a continuous command from the SMC with the AIRD algorithm.

[0280] We can deduce from the figure 22that despite the unbalanced currents and voltages of the network, the significant harmonic voltage distortion of the site studied as well as the presence of the LCL filter, all the control methods ensure near-perfect tracking, reflected, after filtering, by a sinusoidal shape of the current on the electrical network side.

[0281] This can be observed by negligible THD of three-phase currents after filtering.

[0282] Indeed, the ( THD - I s The THDs after filtering are: 0.4%, 0.8%, and 0.5%, these values ​​being the highest of the control methods, respectively: discontinuous (classical) SMC, continuous (sigmoid) SMC, and continuous (AIRD) SMC. It should be noted that the THDs of the currents before activation of the active filter were 11.6%, 10.7%, and 12% for the three phases.

[0283] The significant reduction in the THD of the conducted current, as shown by the Fig. 23 , to a very pronounced reduction in the THD of the voltage at the connection point; the (THD - V s ) go from (6.7%, 6.2% and 6.8%) before filtering to (0.28%, 0.5% and 0.3%) after filtering, respectively for the control methods, discontinuous SMC (classical), continuous SMC (with a Sigmoid function) and continuous SMC (with the AIRD method).

[0284] It should be noted that international standards (e.g., IEEE STD 519-2014) impose a maximum voltage THD of between 5% and 8% in industrial areas, while EDF's recommendations require a total voltage harmonic distortion rate of less than 1.6% for each receiver. Higher-order SMCs

[0285] In this part, we will validate the use of the active filter associated with the LCL filter with the control algorithms, C-HOSM then 2-SMC Twisting and finally 2-SMC Super-Twisting.

[0286] The technical data sheet for the components constituting the active filter is presented via the preceding Table IV.

[0287] THE Figs 24 , 25 , 26 They present the simulation (before and after filtering) of the three-phase line voltages at the Us123 connection points, and the three-phase currents on the network side. I s 123, the three-phase THD of the current (THD - I s ) as well as that of the tension ( THD - V s ) superimposed, using respectively the C-HOSM control algorithm then 2-SMC Twisting and finally 2-SMC Super-Twisting, knowing that in all three cases, the active filter is primed after five periods of the sector (up to 0.1 s).

[0288] We can deduce from these figures that, despite a deteriorated and unbalanced network voltage, as well as the presence of the LCL filter, the three control methods ensure almost perfect tracking, reflected, after filtering, by a sinusoidal form of the current on the electrical network side.

[0289] Indeed, the ( THD - I s The THDs after filtering for the three phases are: (0.92%, 0.65%, 0.75%), (0.55%, 0.35%, 0.42%), and (0.31%, 0.15%, 0.23%) for the C-HOSM, 2-SMC Twisting, and 2-SMC Super-Twisting control algorithms, respectively. It should be noted that the THDs of the currents before activation of the active filter were (11.6%, 10.7%, and 12%) for the three phases, respectively.

[0290] These excellent current filtering results translate into a significant reduction in three-phase voltage THD. (THD - V s) of (6.7%, 6.2% and 6.8%) respectively for the three phases before filtering to (0.59%, 0.55%, 0.66%), (0.23%, 0.16%, 0.25%), (0.2%, 0.15%, 0.22%) after filtering, respectively using the C-HOSM, 2-SMC Twisting and 2-SMC Super-Twisting control algorithms.

[0291] It is worth noting that the three proposed control algorithms preserve the robustness and excellent filtering quality provided by conventional discontinuous SMC, while guaranteeing, through continuous control, a voltage THD well below the 1.6% required by the most demanding recommendation standard (EDF standard). This significant result is achieved despite the unbalanced currents and voltages in the network, the significant voltage harmonic distortion at the studied site, and the presence of the LCL filter.

[0292] It should be noted that international standards (e.g., IEEE STD 519-2014) impose a maximum voltage THD of between 5%-8% in industrial areas, while EDF recommendations require a total voltage harmonic distortion rate of less than 1.6% for each receiver. Calculation and integration of the maximum power point in the control scheme of the active filter:

[0293] The goal here is to track the maximum power point of a photovoltaic generator, in order to increase the efficiency of this generation system. Fig. 27 shows the four parameters characterizing the operation of a PV panel / generator; these are the short-circuit current Icc, the open-circuit voltage Vco, the maximum power current Impp and the maximum power voltage Vmpp (therefore the maximum power available within the PV generator: Pmpp = Vmpp × Impp).

[0294] In this invention, the algorithm used for extracting the maximum power point is the P&O algorithm (in English “Perturb and Observe”), which is based on the disturbance and observation of the voltage of the PV generator, until the maximum voltage is obtained which will correspond to the MPPT point. DC voltage regulation loop (maximum power pursuit):

[0295] We will leverage the capacitor voltage regulation loop on the DC side of the inverter to ensure the PV generator maintains its maximum power output. Indeed, in the case of a parallel active filter not connected to a PV generator, the energy storage capacitor regulates itself (charges while maintaining a constant voltage) via the electrical grid, through the inverter, to compensate for the Joule heating losses of the inverter's power electronics components and the LCL output filter. The capacitor voltage Vdc must track a reference voltage Vdc-ref, the amplitude of which is chosen to enhance the system's dynamics while respecting the sizing requirements of the parallel active filter's electrical components.

[0296] In this invention, the same capacitor voltage regulation loop is used to ensure maximum power point tracking based on the tracking of the power PMPP and not the current IMPP. Indeed, unlike inverter capacitor regulation methods, which adopt the tracking of the current IMPP in the case of a renewable energy generation system, this method directly extracts the maximum power from the PV generator. Modeling the voltage regulation loop of the capacitor:

[0297] The relationship between the active power produced by the PV generator and the voltage across the capacitor can be written in the form: P pv = d dt 1 2 C dc V dc

[0298] Since the relationship (Maths 30) is non-linear, and for small variations of the voltage V dc around its reference V dc-ref, it can be linearized via the following relationship: P pv = C dc V dc − ref d dt V dc

[0299] It should be noted that for: V dc-ref = V MPP (delivered by the P&O algorithm of unit 25-B), we will have P pv = P MPP. Therefore, the capacitor voltage calculated in the Lablace domain becomes: V dc s V dc = V MPP = P MPP s V dc − ref C dc s

[0300] From the relationships (Maths 31) and (Maths 32) and using a linear (Proportional-Integral) or other regulator, the DC voltage regulation loop Vdc and consequently the power P MPP can be presented via the Fig. 28 The choice of parameters for this controller ensures fast and precise tracking of the maximum power output of the PV generator. It should be noted that the measured DC voltage signal will be filtered from fluctuations at 300 Hz or other frequencies via a second-order low-pass filter ( see Fig. 28 ). Integration of the maximum power tracking loop within an active filter:

[0301] The advantage of P MPP tracking (compared to I MPP) is the ability to integrate (with minimal computational effort) the P MPP tracking loop into the algorithm for identifying disturbing currents in the control section of the parallel active filter. In this case, the reference voltage V dc-ref of the capacitor voltage identification / regulation algorithm for the active filter is replaced by the maximum power voltage V MPP (V dc-ref = V MPP) from the (P & O) algorithm, as shown in the diagram. Fig. 29 From this figure 29 Since the output of the PI controller is the maximum power P MPP of the PV generator, this signal is added to the active disturbing power P̃ delivered by the 25-A disturbance current identification algorithm; the instantaneous disturbance powers (reactive Q̃ and homopolar P0) are also calculated. Indeed, this algorithm first calculates the instantaneous disturbing powers caused by non-active disturbing currents (harmonics, reactive and unbalanced, or other) present in the load current IL, in the α, β, and 0 frame. The calculation of the reference currents I ref123 is performed via an inverse process, first calculated in the same α, β, and 0 frame, then in the three-phase frame. These reference currents of the active filter (I ref123) then include the disturbing currents as well as the maximum current of the PV generator: I MPP.It should be noted that by basing on a voltage regulation V dc whose regulator output (PI for example) is the maximum power, we ensure that the instantaneous power method, used in this invention to identify the disturbing currents, remains unchanged while guaranteeing the identification of the maximum current of the PV generator: I MPP in amplitude, with an angle equal to that of the direct voltage component already extracted via the PLL 26 (see . Fig. 2 In this case, the current I MPP will systematically have a phase equal to this angle without passing through the PLL loop, because the voltage used in the 25-A method for identifying non-active disturbance currents is that of the direct component. Therefore, no special identification method, either in amplitude or phase / angle, is required for extracting the current I MPP, as is usually the case in the state of the art.

[0302] Indeed, in addition to the entire section dedicated to identifying disturbing currents that will need to be added, algorithms that adopt current tracking (I MPP) instead of P MPP would involve multiplying I MPP by three functions sinus (phased out by one-third of a period) of unit amplitude and angle / phase from the PLL (angle of the direct component of the electrical network voltage Vs). This requires more calculations and greater precision from the PLL.

Claims

1. Method for modeling the electrical operation of an industrial site comprising a plurality of machines controlled by electronic power converters and distributed across one or more production lines, the controlled machines being modeled by current sources, representing the total consumption (active, reactive, and harmonics) of the controlled machines, measured by an energy analyzer installed at the input of each production line, characterized in that the steps of the method are as follows: 1- installing a single measuring device, which is a spectrum analyzer / power quality analyzer, at the input of each production line of the industrial site under study, each production line comprising several controlled machines, the measuring device providing the electrical quantities of the network: active, reactive, and apparent power; power factor; fundamental components of current and voltage; total harmonic distortion (THD) of current and voltage; individual harmonic distortion of currents and voltages, 2- modeling the active and reactive powers using a parallel R-L branch or a single current source representing the fundamental current, with an angle corresponding to the power factor; the direction of current flow is from the electrical network (1) to a load (2), 3- Modeling the individual harmonic distortion rate of currents and voltages using current sources, each corresponding to a harmonic rank, with the direction of harmonic currents flowing from the load (2) to the electrical network (1). 4- Connecting all these current sources modeled in steps 2 and 3 to one or more resistors of selected values to force the currents to flow in a single direction: - from the electrical network (1) to the loads (2) for the fundamental current, and - from the loads (2) to the electrical network (1) for harmonic currents, and 5- obtaining a model of the electrical operation of the given industrial site (1) in terms of current sources and resistors.

2. Method according to claim 1, in which the network has several voltage levels, comprises several power transformers with cables and power lines, as well as dozens of controlled machines associated with several active pollution control filters.

3. Method according to one of claims 1 to 2, in which the machines are controlled by power converters, such as inverters and rectifiers.

4. Method according to one of claims 1 to 3, in which the given industrial site studied has: - non-linear / linear loads (2), - at least one renewable energy generation power unit (100), and - a renewable energy generation parallel active filter type current compensation device (7), capable of being connected: ➢ at its input, downstream of at least one renewable energy generation power unit (100) coupled to an energy storage element (3), and ➢ in parallel, at its output, upstream of a connection point (C) between, on the one hand, a given electrical network (1) and, on the other hand, nonlinear and linear electrical loads (2), the current compensation device (7) comprising: - a power conversion unit (8), comprising at least one voltage inverter (9) generating an alternating current, with a frequency band ranging from 50 to 2500Hz, covering: a) the entire frequency band of the non-active disturbance current, which has: all or part of the harmonics, and at the fundamental frequency , all or part of the reactive power and / or an imbalance, b) as well as the maximum active current generated by the renewable energy generation power unit (100); - an output filtering unit (10), comprising an LCL-type filter (11) for each of the phases and a neutral, and connected: on the one hand downstream of the voltage inverter (9), and on the other hand in parallel with the connection point (C) between the given electrical network (1) and the non-linear and linear electrical loads (2), the LCL output filter (11) being dimensioned to block the harmonic components due to the switching of the inverter (9); - a control unit (12) comprising a reference current calculation unit (25), the reference currents comprising: i. at least one non-active disturbance current intended to be injected at the connection point in opposite phase to cancel, on the electrical network side (1), the signal disturbances generated by the nonlinear and linear loads (2), presenting all or part of the harmonics, and at the fundamental frequency all or part of reactive and / or unbalanced current, ii. at least one active current for recharging the storage element (3), iii. at least one active current corresponding to a maximum power point of the renewable energy generation power unit (100), the control unit (12) also comprising: a switching control device (21) that controls the switching of the inverter (9) via a nonlinear controller (23) using a continuous sliding mode, adapted to the LCL-type filter (11) which provides closed-loop control of the entire frequency band from 50 to 2500 Hz for the injection of the non-active disturbance current and the active current, based on the identification of the reference currents by the calculation unit (25); the switching control of the inverter (9) being performed in such a way as to allow the output filter unit (10) to pass through the connection point (C): • a part or all the non-active disturbance currents injected in opposition phase comprising harmonic currents, as well as reactive and unbalanced currents at the fundamental frequency, in the nonlinear and linear electrical loads (2), to satisfy the reactive energy consumption demand of the nonlinear and linear electrical loads (2), while cleaning up the electrical network (1) of these reactive disturbance currents; • an active current corresponding to a maximum power point available within the renewable energy generation power unit (100), to satisfy the active energy consumption demand of the nonlinear and linear electrical loads (2), while ensuring the recharging of the storage element (3).

5. Method according to claim 4, wherein the switching control of the inverter (9) is performed so as to also allow the following to pass through the output filter unit (10): a portion of the active current into the given electrical network (1), which is then free of non-active disturbance currents, when the output of the renewable energy generation power unit (100) exceeds the power consumed by the nonlinear and linear electrical loads (2).

6. Method according to one of claims 4 or 5, in which the renewable energy generation power unit (100) is coupled without a chopper or other power electronic devices to the energy storage element (3).

7. Method according to any of claims 4 to 6, wherein the control unit (12) comprises: - a chopper configured to maintain a predefined constant DC voltage across the energy storage element (3) of the inverter (9), ly of the voltage level of the renewable energy source (100) to ensure unchanged harmonic filtering. - a double chopper for the case of an isolated electrical network.

8. Method according to any of claims 4 to 7, in which the control unit (12) comprises a continuous sliding mode controller of1storder or higher order, having a sliding surface adapted to1storder,2ndorder,3rdorder, or others.

9. Method according to any of claims 4 to 8, wherein the control unit (12) comprises at least one of the following two sliding mode controllers: - one comprising an approximate sign function as a sigmoid function with a 2nd-order sliding surface; and - the other is based on an artificial relative degree increase method comprising a sign function with an adapted3rdorder sliding surface, followed by an integrator.

10. Method according to any of claims 4 to 9, in which the control unit (12) includes at least one of the following three higher-order sliding mode controllers: - the first is based on a continuous higher-order sliding mode controller C-HOSM comprising three sign functions, with an adapted sliding surface of the 1st order, - the second uses a 2-SMC Twisting controller with artificial increase in relative degree, comprising two Sign functions, with an adapted sliding surface of 2nd order, followed by an integrator, - the third is a 2-SMC Super-Twisting controller with a Sign function, with an adapted sliding surface of the 2nd order, without artificial increase in the relative degree.

11. Method according to any of claims 4 to 10, in which the renewable energy generation power unit (100) is: - selected from the following list: one or more photovoltaic panels, wind turbine(s), fuel cell(s) or others, - coupled directly to a capacitive storage element (3) in the case of continuous production, or via an AC / DC power rectifier in the case of alternating production.

12. Method according to any of claims 4 to 11, wherein the grid is selected from the following list: the main electrical grid, a local microgrid that is isolated or connected to the main electrical grid, or an on-board electrical grid.

13. Method according to any of claims 4 to 11, further comprising a smart building (27), and in which the control unit (12) is connected to a decentralized management unit (70) of the smart building (27), the control unit (12) compares the maximum available power of the renewable energy generation unit (100) with the total load of the smart building (27).

14. Method according to claim 13, wherein the control unit (12) is configured to optimize the consumption of the various devices operating within this smart building (27) by distributing the loads corresponding to the non-linear / linear loads according to at least two operating modes: - a first distribution mode known as the adapted consumption mode, in which the control unit (12) controls the decentralized management unit (70) of the smart building (27) so as to adapt the consumption of the smart building to the production of the renewable energy generation power unit (100), so that the total load curve of the smart building has a maximum simultaneity factor corresponding to the operation of all the building's loads at the same time, within the limits of the renewable energy produced, - in the event of insufficient renewable energy production, a second distribution mode known as the modulated consumption mode, in which the control unit (12) controls the decentralized management unit (70) of the smart building (27) in order to modulate the consumption of the smart building's appliances (27) to achieve a total load curve for the smart building that is substantially constant over time, - in the same scenario and if the electrical network (1) is smart, a third distribution mode adapted to the production of the smart electrical network (1) is possible.

15. Method according to any of claims 10 to 14, further comprising: - a local network (28) connected to the network (1), and - conventional (fossil) power generation units (80), and - a semi-decentralized management system (29), and - a plurality of renewable energy generation power units (100) connected to the local network (28) by a compensation device (7), each compensation device (7) being connected to the semi-decentralized management system (29) to which it communicates information concerning the current and future energy production of each of the renewable energy generation power units (100), and - a plurality of consumption stations corresponding to non-linear and linear loads (2), each of the consumption stations being connected to the local network (28) and equipped with a compensation device (7), connected to the semi-decentralized management system (29) to which it communicates information concerning instantaneous consumption and future consumption based on the programmed operation of the consumption stations, - a plurality of renewable energy generation power units (100) and linear and non-linear loads (2) of positive energy smart buildings (27), which ensure self-consumption and where any surplus energy is stored or exchanged with other smart buildings or delivered to the local grid (28) via the control unit (12) of the compensation device (7), in coordination with the semi-decentralized management system (29) in the event of exchange with the local grid (28).