TWI462419B - Neural static switch - Google Patents

Description

Neurostatic static switch

The present invention relates to a neurological static switch, and more particularly to a static switch that combines wavelet transform processing and an Adaptive Network-Based Fuzzy Inference System (ANFIS) module to control switch disconnection and connection. .

In recent years, the concentration of greenhouse gases in the atmosphere has reached record highs. The World Meteorological Organization (WMO) reported in 2011 that 13 years have been the warmest year in history in the past 15 years. In addition to causing the sea level to rise, the warming phenomenon causes the ecological imbalance of the species to bring about rapid climate change. In the face of the severe test brought about by climate change, the developed countries in the world have planned to gradually achieve greenhouse gas zero between 2050 and 2060. The target of emissions, as for developing countries, also hopes to achieve zero growth in greenhouse gas emissions by 2020.

Among them, with the rapid changes in the climate, Taiwan in the summer is bound to become a victim of this wave of climate warming. In addition to the climatic disadvantages, Taiwan is located in the seismic belt around the Pacific Ocean, which is a typical plate collision. The islands produced have become countries with frequent earthquakes. In the case of Japan's 311 earthquake, the Tokyo Electric Power Company's Fukushima nuclear power plant caused a large-scale power outage due to the tsunami caused by the earthquake, causing a large-scale power outage. As a result, one-third of the power comes from the current power generation. Nuclear power generation in Japan is still implementing power-saving measures, causing direct rush to the people and industry. hit.

Taiwan, like Japan, adopts Centralized Generation (CG), which is dominated by thermal power plants, nuclear power plants and hydroelectric power plants. In the event of a natural disaster, an earthquake or improper human operation, a large-scale power outage will occur, which will cause serious losses to Taiwan, which is heavily targeted by the technology industry, and thus impact Taiwan’s economy and competitiveness. Therefore, a micro-grid with energy-saving and environmental protection issues and a regional-based approach will emerge.

Microgrids are composed of systems such as loads, static switches, decentralized power supplies, power equipment, and automatic monitoring. Dispersed power supplies are more common in solar, wind, fuel cell, and micro-turbine generator systems. Microgrids are generally connected in the distribution system. In order to avoid the power flow change caused by the current flow in the power system, power is usually not supplied into the distribution network. When a fault occurs in the power distribution grid or the microgrid, the static power switch detects the microgrid and the power distribution system, and runs in the island operation mode to avoid the disturbance caused by the faulty party and keep the system stable. .

However, the existing static switch is still mainly mechanical, and it needs to be switched by manpower after detecting a fault. In addition, the mechanical static switch occupies a considerable space and its construction is large, so the existing The static switch still needs to be improved.

In view of the prior art, the existing static switch is still a mechanical static switch, which occupies a large space and is relatively expensive to construct. Some static switches still need to be improved. Accordingly, the main object of the present invention is to provide a neural-like static switch which is controlled by wavelet transform processing and adaptive network-based fuzzy inference system (ANFIS) module. The switch is disconnected and connected to achieve automatic switch switching.

The present invention solves the problems of the prior art, and the necessary technical means is to provide a neural-like static switch for connecting to a micro-grid system and a mass power supply system, and the micro-grid system includes at least one power generation area system. The mass power supply system is electrically connected to the neural-like static switch, and the power generation area system is electrically connected to the neuro-static static switch via a main bus bar, and includes at least one power generation module, and the neural-like static switch includes A first wavelet transform module, a second wavelet transform module, at least one Adaptive Network-Based Fuzzy Inference System (ANFIS) module, a processing module and a switch. The first wavelet transform module is configured to detect a voltage of a main bus bar for performing a wavelet transform process according to the voltage condition, thereby outputting a first detection signal. The second wavelet conversion module is configured to detect a current of the main bus, and perform wavelet conversion processing according to the current condition to output a second detection signal.

The ANFIS module is electrically connected to the second wavelet transform module for receiving and processing the second detection signal, thereby outputting a type of neural processing signal. The processing module is electrically connected to the first wavelet transform module and the ANFIS module The group is configured to receive the first detection signal and the neural-like processing signal, thereby outputting a control signal according to the first detection signal and the neural-like processing signal. The open relationship is electrically connected to the processing module for switching according to the control signal, so that the microgrid system is connected or disconnected from the mass supply system.

Preferably, in the above-mentioned neural static switch, the ANFIS module includes at least one exponent address processor and at least one look-up table processor for processing the second detection signal. In addition, the first wavelet transform module and the second wavelet transform module comprise a plurality of signal processors, wherein the signal processors are a converter, an adder, and a multiplier; the first wavelet transform module and the second The wavelet transform module includes a plurality of low pass filters electrically connected to the signal processors and a high pass filter. In addition, a signal processing module is included, and the signal processing module is electrically connected to the second wavelet transform module and the ANFIS module for performing a bussew processing on the second detection signal.

Preferably, in the above-mentioned neurostatic static switch, the processing module further comprises an OR logic circuit and an AND logic circuit, the OR logic circuit is electrically connected to the ANFIS module, and the AND logic circuit is Electrically connected to the first wavelet transform module and the OR logic circuit, the OR logic circuit receives the second detection signal and outputs a logic signal to the AND logic circuit, so that the AND logic circuit is based on the first detection signal and the logic signal. Output control signals. In addition, the power generation module is one of a solar power generation module, a wind power generation module, and a micro turbine power generation module.

Compared with the prior art, the neural static open relationship provided by the present invention is automatically processed through the wavelet transform processing and the adaptive network-based fuzzy inference system (ANFIS) module. Controlling the effect of the switch being disconnected and connected, and the above processing is realized by the wafer, thereby improving the conventional mechanical static switch into a wafer type static switch, thereby further reducing the occupied space and the construction cost, and The accuracy of automatically shutting down when detecting a fault is quite high, thus solving the problems of the prior art.

The specific embodiments of the present invention will be further described by the following examples and drawings.

Since the combination of the embodiments of the neural static switch provided by the present invention is numerous, it will not be repeated here, and only a preferred embodiment will be specifically described.

Please refer to FIG. 1 and FIG. 1A together. The first figure shows a first block diagram of a neural static switch connected to a mass public power system and a micro grid system according to a preferred embodiment of the present invention. A second block diagram showing a neural static switch of the preferred embodiment of the present invention coupled to a mass public electrical system and a microgrid system is shown. As shown, the neuro-static static switch 1 is used to connect to a micro-grid system 2 and a mass supply system 3, the micro-grid system 2 includes three power generation regional systems 21, 22, 23, and the mass-powered system 3 is electrically Connected to the neuro-static static switch 1, and the power generation area systems 21, 22, 23 are electrically connected to each other via a main bus bar 4 The neurostatic static switch 1 and the power generation area system 21, 22, 23 comprise at least one power generation module.

The power generation module is, for example, a solar power generation module, a wind power generation module, or a micro turbine power generation module. In other embodiments, for example, a tidal power generation module or a geothermal power generation module can generate environmentally friendly power generation. Module. In a preferred embodiment of the present invention, the power generation area system 21 includes a solar power generation module 211 and a micro-turbine power generation module 212; the power generation area system 22 includes a solar power generation module 221 and wind power generation modules 222, 223; The power generation area system 23 includes a solar power generation module 231 and a micro turbine power generation module 232. Of course, it is not limited to the above configuration in other embodiments. In addition, the Volkswagen Power Supply System 3 is a power supply system consisting of the Taiwan Power Company (Taiwan Power).

Specifically, please refer to FIG. 1A, which is a schematic diagram of power distribution in practice. The Volkswagen power supply system 3 is electrically connected to the neuro-static static switch 1 via the bus bar 5 and the bus bar 6, respectively, and the mass supply is The voltage value supplied by the system 3 is, for example, 69 kV, and when it passes through the bus bar 5, it is transformed into a 11.4 kV via a transformer (not shown), and when passing through the bus bar 6, it is further passed through a transformer (not shown). The voltage is 380V, that is, the voltage of the bus bar 5 is 11.4kV, and the voltage of the bus bar 6 is 380V, wherein the above is only an example, and other values may be other embodiments. Depending on the national power standards, it is hereby stated.

The solar power generation module 211 and the micro-turbine power generation module 212 of the power generation area system 21 are electrically connected to the neuro-static static switch 1 via the bus bar 7 and via the main bus bar 4; The power generation module 221 and the wind power generation modules 222 and 223 are electrically connected to the neuro-static static switch 1 via the bus bar 9 and via the bus bar 8 and the main bus bar 4; the solar power generation module 231 of the power generation area system 23 And the micro-turbine power generation module 232 is electrically connected to the neuro-static static switch 1 via the bus bar 10 and then via the bus bar 8 and the main bus bar 4, wherein, if the voltage value is continued, the main bus bar 4 is continued. The voltage values of the bus bars 7, 8, 9, and 10 are both 380 V. Of course, this configuration varies depending on the actual adjustment, and is not limited to the above.

Please refer to the second to fourth B drawings. The second figure shows a block diagram of a neural static switch according to a preferred embodiment of the present invention, and the third figure shows the first wavelet transform of the preferred embodiment of the present invention. 4 is a schematic diagram showing an ANFIS module according to a preferred embodiment of the present invention, and FIG. 4A is a schematic diagram showing an exponential address processor according to a preferred embodiment of the present invention, and FIG. 4B is a schematic diagram showing A schematic diagram of a look-up table processor in accordance with a preferred embodiment of the present invention.

As shown, the neuro-static static switch 1 includes a first wavelet transform module 11, a second wavelet transform module 12, a signal processing module 13, and a four adaptive neuro-fuzzy inference system (Adaptive Network-Based). Fuzzy Inference System; ANFIS) modules 14, 14a, 14b and 14c, processing module 15 and switch 16. The first wavelet transform module 11 is electrically connected to the main bus bar 4, the second wavelet transform module 12 is also electrically connected to the main bus bar 4, and the signal processing module 13 is electrically connected to the second wavelet transform module. 12.

The internal circuit structures of the first wavelet transform module 11 and the second wavelet transform module 12 are the same. Here, only the internal circuit structure of the first wavelet transform module 11 is taken as an example. Please refer to the third figure. A wavelet conversion module 11 includes a plurality of signal processors 111 and a plurality of low pass filters 112. In the preferred embodiment of the present invention, the signal processor 111 included in the first wavelet transform module 11 has a converter (Z ^{-1 in the} upper row), a multiplier (a*b in the middle row), and an adder (in the lower row). The low-pass filter 112 is electrically connected to the signal processor 111 described above, and is used as an input of a multiplier in the signal processor 111, and the output of the multiplier is used as the signal processor 111. The input of the adder, wherein the internal circuit structure of the first wavelet transform module 11 is only one of the embodiments, which is based on the actual adjustment, and thus is not limited to the above and shown in the figure.

For example, in other embodiments, the low pass filter 112 described above may be replaced by a high pass filter due to practical adjustments. Specifically, as shown in the following table, the filter in the preferred embodiment of the present invention The type of the device can be transformed by the adjustment means of the coefficients. The following sequence 0 to 7 represent the left-to-right filter of the third figure.

The ANFIS modules 14, 14a, 14b and 14c are electrically connected to the signal processing module 13, wherein the ANFIS modules 14, 14a, 14b and 14c respectively simulate the mass supply system 3 (outside the network), the power generation area system 21, and the power generation. The regional system 22 and the power generation area system 23, and the internal circuit structures of the ANFIS modules 14, 14a, 14b, and 14c are the same. In the preferred embodiment of the present invention, only the internal circuit structure of the ANFIS module 14 is taken as an example. In the four-figure, the ANFIS module 14 includes a three-index address processor 141 (only one is shown), a three-table processor 142 (only one is shown), a rule processing module 143, and a conclusion inference processing module. 144, and an output processing module 145, the look-up table processor 142 is electrically connected to the exponent address processor 141, the rule processing module 143 is electrically connected to the look-up table processor 142, the conclusion inference processing module 144 The system is electrically connected to the rule processing module 143, and the output processing module 145 is electrically connected to the conclusion inference processing module 144.

The ANFIS module 14 is a module composed of an adaptive neuro-fuzzy theory, which mainly includes an input layer, a rule layer, a conclusion inference layer, and an output layer. Specifically, the exponent address processor 141 and the lookup table The processor 142 is the input layer, the rule processing module 143 is a rule layer, the conclusion inference processing module 144 is a conclusion inference layer, the output processing module 145 is an output layer, and the exponent address processor 141, The lookup table processor 142, the rule processing module 143, the conclusion inference processing module 144, and the output processing module 145 are all implemented by logic circuits. In addition, it is worth mentioning that the ANFIS module 14 of the preferred embodiment of the present invention does not include The normalization layer included in the traditional adaptive neuro-fuzzy theory is hereby described.

For further reference to FIG. 4A, in the exponent address processor 141 of the preferred embodiment of the present invention, the exponent address processor 141 includes a subtractor 14101, a subtractor 14102, a comparator 14103, and a comparator. 14104, a multiplexer 14105, a multiplexer 14106, a divider 14107, a multiplier 14108, a converter 14109, and a subtractor 14110. The subtracter 14101 is configured to receive an input of "x_end" and an "x_start"; the subtractor 14102 is electrically connected to the multiplexer 14106, and the comparator 14103 is electrically connected to the multiplexer 14105, and is configured to receive the "x_end" and an input of "x"; the comparator 14104 is electrically connected to the multiplexer 14106 and is configured to receive the input of the "x_start" and the "x"; the multiplexer 14106 is configured to receive the "" The output of x_start", comparator 14104, and multiplexer 14105 is input; the divider 14107 is electrically coupled to the subtractors 14101, 14102 and has its output as an input; the multiplier 14108 is electrically coupled to the divider 14107, and The converter 14109 is electrically connected to the multiplier 14108 and receives its output as an input, and its output is "index_int"; the subtractor 14110 is electrically connected to the multiplier. 14108 and a converter 14109 for receiving an output thereof as an input and outputting "index_dec", wherein the above circuit structure is only the structure of the preferred embodiment of the present invention, and other embodiments may be other junctions. This is hereby stated.

For further reference to FIG. 4B, in the look-up table processor 142 of the preferred embodiment of the present invention, the look-up table processor 142 includes a memory 14201, A comparator 14202, an adder 14203, a multiplexer 14204, a memory 14205, a subtractor 14206, an adder 14207, a comparator 14208, a multiplier 14209, and a multiplexer 14210. The storage unit 14201 is configured to receive the input of the “index_int”, and the storage unit 14201 stores a lookup table correspondence table, and the lookup table correspondence table includes the correspondence between the input data and the attribution function, and the storage 14201 The comparator 14202 is configured to receive the input of the "index_int" and a constant "255"; the adder 14203 is configured to receive the input of the "index_int" and a constant "1"; The multiplexer 14204 is electrically connected to the comparator 14202 and the adder 14203, and is configured to receive the output and the constant "255" as an input; the storage 14205 is electrically connected to the multiplexer 14204 for receiving the output thereof. As an input, the storage 14205 stores a lookup table correspondence table, and the lookup table correspondence table includes the correspondence between the input data and the attribution function, and the storage 14205 can be a memory having a memory function generally; the subtractor The 14206 is electrically connected to the storage 14201 and the storage 14205 for receiving the output thereof as an input; the adder 14207 is electrically connected to the storage 14201 And a multiplier 14209 for receiving its output as an input; a comparator 14208 is electrically coupled to the subtractor 14206 for receiving its output and a constant "0" as an input; the multiplier 14209 is electrically coupled to the subtractor 14206 The multiplexer 14210 is electrically connected to the memory 14201, the adder 14207, and the comparator 14208 for receiving the output as an input, and the output thereof is "". MF", wherein the above circuit structure is only a preferred embodiment of the present invention The structure of the example can be other structures in other embodiments, and is hereby described.

In addition, please continue to refer to the second to fourth B diagrams, the first wavelet conversion module 11 is used to detect the voltage of one of the main bus bars 4, thereby performing a wavelet conversion process according to the voltage condition. The first detection signal S1 is output, and the second wavelet conversion module is configured to detect a current of the main bus 4 for performing wavelet conversion processing according to the current condition, thereby outputting a second detection. Signal S2. Specifically, the wavelet transform processing described above reduces the complexity of the original continuous wavelet transform (CWT) calculus by a discrete wavelet transform (DWT) calculus, and the preferred embodiment of the present invention The wavelet transform processing decomposes the data of the voltage and current into three layers (order) of high frequency signals via the architecture as shown in the third figure, that is, the first detection signal S1 and the second detection signal S2. Contains three layers of high frequency signals. However, the practice can also be decomposed into four layers, five layers or even more, and thus is not limited to the above three layers. In addition, the first detection signal S1 and the second detection signal S2 are digital logic signals, which are “1” or “0”. For example, in the preferred embodiment of the present invention, When the voltage and current of the main bus 4 are abnormal, the first detection signal S1 and the second detection signal S2 are "1", and the normal is "0".

The signal processing module 13 is configured to perform a Baccarat process on the second detection signal S2. Specifically, the Bassava process processes the second detection signal S2 according to the Basseva energy theorem. Knowing the relationship between the energy of the wavelet coefficients of each layer in the second detection signal S2, that is, the energy division of the signal in the time domain and the frequency domain of the wavelet transform processing, and thus utilizing the small The energy characteristic values of the first to third order waves are used as the basis for identifying signals of normal or abnormal. If you collect the normal or abnormal one-to-third-order energy feature data and the normal and abnormal occurrence points mapped by these data, you can use these data to form ANFIS training and learn the fuzzy attribution of the first to third-order wavelet energy features. The parameters of the function.

The ANFIS modules 14, 14a, 14b and 14c are configured to receive and process the second detection signal S2, thereby outputting a type of neural processing signals S3, S3a, S3b and S3c, in particular, via the exponent address processor 141. For the attribution function included in the second detection signal S2, the exponential address is obtained by interpolation and a plurality of approximation methods, and as shown in FIG. 4A, the "x" table ANFIS modules 14, 14a, 14b With the input of 14c, "x_end" and "x_start" indicate the upper and lower limits, respectively. In the first step of the calculation, the comparator 14103 and the comparator 14104 determine whether the "x" exceeds the upper and lower limits, and divide the difference between the "x" and the "x_start" calculated by the subtractor 14101 by the subtractor. 14102 calculates the difference between "x_end" and "x_start", and further knows that the index corresponds to the address of the look-up table method, and in the preferred embodiment of the present invention, the above-mentioned table look-up method is divided into 256 copies. Therefore, the output of the divider 14107 is multiplied by "256" by the multiplier 14108, and finally the index address is divided into integers and decimals via the converter 14109 and the subtractor 14110, that is, the "index_int" of the fourth A picture is an integer. , and "index_dec" is a decimal.

After the above result is generated, it is input to the look-up table processor 142. Further, it is implemented by the look-up table correspondence table built in the memory 14201, and the attribution function is obtained from the corresponding relationship. First of all, It is first determined by the comparator 14202 whether "index_int" is the upper limit value. If it is the upper limit value, the result of the look-up table is directly output by the memory 14201; if it is not the upper limit value, the "index_int" is added by the adder 14203. After adding "1", it is input to the memory 14205 via the multiplexer 14204 to obtain the result of the lookup of the next address, and then the result is compared with the output of the memory 14201 via the subtractor 14206. The difference is obtained, and finally the multiplier 14209 is multiplied by the difference value and "index_dec", and then added to the output of the memory 14201 to obtain the attribution function "MF". In addition, it is worth mentioning that the index address processor 141 and the look-up table processor 142 in the ANFIS module 14 in the fourth figure each have three, which are mainly after the wavelet decomposition, the second detection signal has three The layer high frequency signals are respectively input to the three index address processor 141, that is, the first layer, the second layer, and the third layer are respectively input to the three index address processors 141.

Then, the neural processing signals S3, S3a, S3b, and S3c are output to the processing module 15 via the processed signals through the processing of the rule processing module 143, the conclusion inference processing module 144, and the output processing module 145. In the preferred embodiment of the present invention, the neural-like processing signals S3, S3a, S3b, and S3c are signals of "1" or "0", which are in the rule processing module 143, the conclusion inference processing module 144, and the output. During the processing of the processing module 145, the signal of "1" or "0" is generated by the actual explicit calculation. In other embodiments, the signal may be generated by comparison, for example, greater than a preset. The threshold sets the output to "1" and vice versa.

Wherein the OR logic circuit 151 receives the above-mentioned neural processing signal After S3, S3a, S3b and S3c, the logic signal S4 is immediately output, and the logic signal S4 is also "1" or "0", and the sum logic circuit 152 is controlled according to the first detection signal S1 and the logic signal S4. The signal S5 is used to switch the switch 16 according to the control signal S5, so that the microgrid system 2 is connected or disconnected to the mass supply system 3. For example, if the voltage of the main bus 4 is abnormal and the fault is abnormal, the first detection signal S1 is 1, and the power generation area system 22 simulated by the ANFIS module 14b is abnormal, and it is in the ANFIS. When the modules 14, 14a and 14c are normal, the logic circuit 151 outputs a logic signal S4 of "1". At this time, the logic circuit 152 outputs a control signal S5 of "1", thereby making the switch 16 known. The abnormal condition occurs and is disconnected from the mass power supply system 3. Further, the power generation area system 22 is further selected to be disconnected from the mass power supply system 3, and only the remaining power generation area systems 21 and 23 are connected to the mass supply system 3.

In order to make the description of the present invention clearer, please refer to the first A figure and the fifth figure to the fifth I picture. The fifth figure to the fifth I figure show the simulation of the neural static switch of the preferred embodiment of the present invention. The result is schematic. As shown in the figure, in the simulation situation of the preferred embodiment of the present invention, the wind speed is set to 9 m/s (the wind power generation module 222 is about 11 kW, the wind power generation module 223 is about 1.5 kW), and the sunshine intensity is 600 W/ m ² (the solar power generation module 211 of the power generation area system 21 is about 17 kW, the solar power generation module 221 of the power generation area system 22 is about 6 kW, and the solar power generation module 231 of the power generation area system 23 is about 30 kW), and the power generation area system The micro-turbine power generation module 212 of 21 outputs a fixed power of 37 kW, and the total load (not shown) is 100% (150 kW). The micro-grid system 2 adopts the master-slave control method (known technology, no As described again, assuming that a symmetrical three-phase short-circuit fault occurs in the busbar 7 at 2 seconds, the neuro-static static switch 1 detects a fault (which is described above), and then powers the Volkswagen after 1/4 cycle. The system 3 is disconnected, and the busbar 7 electrically connected to the power generation area system 21 is disconnected and disconnected, so that the solar power generation module 211 and the micro turbine power generation module 212 are disconnected from the load (not shown), thereby The microgrid system 2 is switched from the parallel mass supply system 3 mode to the island mode ( Known technology, no longer described in detail, and further, the micro-turbine power generation module 232 of the power generation area system 23 is used as the main power source, and the converter (not shown) adopts voltage-frequency control, and the remaining solar power generation modules 221, The wind power generation modules 222 and 223 and the solar power generation module 231 adopt PQ control (known techniques, and will not be described again).

Wherein, as shown in FIG. 5 to FIG. 5B, the high frequency signals of the first layer, the second layer and the third layer of the main bus bar 4 voltage of the microgrid system 2 after the wavelet conversion process are apparently It can be found that the fifth A picture is most obvious for detecting the symmetrical three-phase short circuit fault of the main bus bar 4. It is worth mentioning that in the preferred embodiment of the present invention, the above obvious contrast is the first layer, The second layer and the third layer are further compared with the high-frequency signal waveform diagram. It can be seen that the amplitude of the second layer is relatively small, so that it is most obvious. Of course, in other embodiments, if the amplitude is compared with the other two If the figure is large, it can be judged more obviously, and it is hereby stated.

Further, the fifth C diagram shows the output power of the microturbine power generation module 232 of the power generation area system 23; the fifth D diagram shows the output power of the micro turbine power generation module 212 of the power generation area system 21; The output power of the wind power generation modules 222 and 223 is represented, and the output power of the wind power generation module 222 is represented by the curve 100, the output power of the wind power generation module 223 is represented by the curve 200, and the solar power generation module 211 is represented by the fifth F diagram. The output power of 221 and 231, the curve 300 is the output power of the solar power generation module 231 of the power generation area system 23, the curve 400 is the output power of the solar power generation module 211 of the power generation area system 21, and the curve 500 is the power generation area system 22 The output power of the solar power generation module 221; the fifth G diagram represents the voltage of the main busbar 4; the fifth H diagram represents the frequency of the main busbar 4; and the fifth diagram I represents the current of the main busbar 4.

Wherein, the microturbine power generation module 232 of the power generation area system 23 controls the output power by 37 kW before the symmetrical three-phase short-circuit fault occurs in the bus bar 7, and when the fault occurs, the voltage-frequency control is adopted, and the bus line 7 is cut off. After that, the total power generation of the other solar power generation modules 221, the wind power generation modules 222, 223, and the solar power generation module 231 is about 48.5 kW, the total load is 90 kW, and power generation is maintained in order to maintain the balance between supply and demand in the microgrid system 2. The output power of the microturbine power generation module 232 of the regional system 23 is adjusted to 45 kW.

Accordingly, it can be seen from the above and the fifth to fifth figures that when the microgrid system 2 is switched to the island operation mode by the parallel mass supply system 3, the power generation module in the microgrid system 2 can still be stably operated and continuously given. Power, and when other power generation modules are tripped, the power generation area system 23 as the main power source can immediately adjust the output power to maintain balance.

In combination with the above, combined with the wavelet static conversion module and the neural static switch such as the ANFIS module, it can infer the faulty power generation area system. The wavelet can provide the relationship between time and frequency and the relationship between time and size, so it can effectively reflect the fault signal. In addition, the adaptive neuro-fuzzy theory applied by the ANFIS module has self-learning ability and can directly process qualitative knowledge and logical inference. In addition, it is based on the wavelet energy characteristic as input, and the result of the above simulation shows that it has a relatively high accuracy, and it is known from the simulation result that after the 1/4 cycle of the fault occurs, the neural-like static switch is immediately disconnected, and At this time, the microgrid system 2 still maintains a stable voltage and frequency, and provides a good load power quality.

Therefore, by the neural static switch provided by the present invention, the effect of automatically controlling the opening and disconnecting of the switch is achieved through the wavelet transform processing and the processing of the ANFIS module, and the above processing is performed by the field programmable logic gate. The implementation of a Field Programmable Gate Array (FPGA) chip, which in turn improves the conventional mechanical static switch into a chip-type static switch, thereby further reducing the occupied space and the construction cost, and replacing it with a neuron type such as a wafer type. After the static switch, the accuracy of detecting the fault or not and automatically closing is still quite high, thus solving the problems of the prior art.

The features and spirit of the present invention will be more apparent from the detailed description of the preferred embodiments. On the contrary, the intention is to cover various modifications and equivalents within the scope of the invention as claimed.

1‧‧‧-type neural static switch

11‧‧‧First Wavelet Conversion Module

111‧‧‧Signal Processor

112‧‧‧Low-pass filter

12‧‧‧Second wavelet transform module

13‧‧‧Signal Processing Module

14, 14a, 14b, 14c‧‧‧ ANFIS modules

141‧‧‧index address processor

14101, 14102‧‧‧ subtractor

14103, 14104‧‧‧ comparator

14105, 14106‧‧‧Multiplexer

14107‧‧‧ divider

14108‧‧‧Multiplier

14109‧‧‧ converter

14110‧‧‧Subtractor

142‧‧ ‧ lookup processor

14201‧‧‧Storage

14202‧‧‧ comparator

14203‧‧‧Adder

14204‧‧‧Multiplexer

14205‧‧‧Storage

14206‧‧‧Subtractor

14207‧‧‧Adder

14208‧‧‧ comparator

14209‧‧‧Multiplier

14210‧‧‧Multiplexer

143‧‧‧ rule processing module

144‧‧‧Conclusion inference processing module

145‧‧‧Output processing module

15‧‧‧Processing module

151‧‧‧ or logic circuit

152‧‧‧ and logic circuits

16‧‧‧ switch

2‧‧‧Microgrid system

21‧‧‧Power Generation Area System

211‧‧‧Solar power module

212‧‧‧Micro Turbine Power Generation Module

22‧‧‧Power Generation Area System

221‧‧‧Solar power module

222, 223‧‧‧ wind power generation module

23‧‧‧Power Generation Area System

231‧‧‧Solar power module

232‧‧‧Micro Turbine Power Generation Module

3‧‧‧VW Power Supply System

4‧‧‧Main bus

5, 6, 7, 8, 9, 10‧‧ ‧ busbars

100, 200, 300, 400, 500‧‧‧ curve

S1‧‧‧ first detection signal

S2‧‧‧ second detection signal

S3, S3a, S3b, S3c‧‧‧ nerve processing signals

S4‧‧‧ logic signal

S5‧‧‧ control signal

The first figure shows a first block diagram of a neural static switch connected to a mass public power system and a micro grid system according to a preferred embodiment of the present invention; the first A figure shows a neural static switch according to a preferred embodiment of the present invention. A second block diagram of a public electrical system and a microgrid system; a second block diagram showing a neural static switch of a preferred embodiment of the present invention; and a third diagram showing the first embodiment of the preferred embodiment of the present invention. A schematic diagram of a wavelet transform module; a fourth diagram showing a schematic diagram of an ANFIS module in accordance with a preferred embodiment of the present invention; and a fourth diagram showing a schematic diagram of an index address processor in accordance with a preferred embodiment of the present invention; A schematic diagram showing a look-up table processor of a preferred embodiment of the present invention; and fifth through fifth figures are diagrams showing simulation results of a neural static switch according to a preferred embodiment of the present invention.

1‧‧‧-type neural static switch

11‧‧‧First Wavelet Conversion Module

12‧‧‧Second wavelet transform module

13‧‧‧Signal Processing Module

14, 14a, 14b, 14c‧‧‧ ANFIS modules

15‧‧‧Processing module

151‧‧‧ or logic circuit

152‧‧‧ and logic circuits

16‧‧‧ switch

4‧‧‧Main bus

S1‧‧‧ first detection signal

S2‧‧‧ second detection signal

S3, S3a, S3b, S3c‧‧‧ nerve processing signals

S4‧‧‧ logic signal

S5‧‧‧ control signal

Claims (7)

A neuro-static static switch is used for connecting to a micro-grid system and a mass supply system, the micro-grid system comprising at least one power generation system electrically connected to the neural static switch. The power generation area system is electrically connected to the neural static switch via a main bus bar, and includes at least one power generation module. The neural static switch includes: a first wavelet conversion module, which is used for Detecting a voltage of the main bus to perform a wavelet conversion process according to the condition of the voltage, thereby outputting a first detection signal; and a second wavelet conversion module for detecting the main convergence The current of the current is arranged to perform the wavelet transform processing according to the current condition, thereby outputting a second detection signal; at least one Adaptive Network-Based Fuzzy Inference System (ANFIS) module The group is electrically connected to the second wavelet transform module for receiving and processing the second detection signal, thereby outputting a type of neural processing signal; The module is electrically connected to the first wavelet transform module and the adaptive neuro-fuzzy inference system module, for receiving the first detection signal and the neural processing signal, according to the first detection The measurement signal and the neural processing signal output a control signal; and a switch electrically connected to the processing module for switching according to the control signal, so that the micro grid system is connected or disconnected from the mass supply system System. The neural static switch of claim 1, wherein the ANFIS module comprises at least one index address processor and at least one look-up table processor for processing the second detection signal. The neural static switch of the first aspect of the invention, wherein the first wavelet transform module and the second wavelet transform module comprise a plurality of signal processors, and the plurality of electrical connectors are electrically connected to the plurality of One of a low pass filter of the signal processor and a high pass filter. A neural static switch as described in claim 3, wherein the signal processors are a converter, an adder, and a multiplier. The neural static switch of claim 1, further comprising a signal processing module electrically connected to the second wavelet transform module and the ANFIS module. Performing a Bacsay process on the second detection signal. The neural static switch as described in claim 1, wherein the processing module further comprises an OR logic circuit and an AND logic circuit, wherein the OR logic circuit is electrically connected to the An ANFIS module, wherein the AND logic circuit is electrically connected to the first wavelet conversion module and the OR logic circuit, and the OR logic circuit receives the second detection signal and inputs And generating a logic signal to the AND logic circuit, so that the AND logic circuit outputs the control signal according to the first detection signal and the logic signal. The neural static switch as described in claim 1, wherein the power generation module is one of a solar power generation module, a wind power generation module, and a micro turbine power generation module.