A pulse capacitor series-parallel switching system based on the Monte Carlo method

Abstract

The present invention discloses a pulse capacitor series-parallel switching system based on the Monte Carlo method, which includes a pulse combination circuit. The pulse combination circuit includes a plurality of pulse circuits directly or indirectly connected to each other; each pulse circuit includes a pulse capacitor and a power switch connected in series; each power switch is controlled to be turned on or off by a control signal to change the connection mode of each pulse circuit in the pulse combination circuit; the control signal is generated by a controller according to the required output voltage and / or total capacitance capacity by executing the Monte Carlo search algorithm. This solution uses the Monte Carlo method to achieve flexible control of the capacitor network, solves the technical problem of frequently changing the capacitance in vehicle relay testing, can realize arbitrary series-parallel switching of the pulse capacitor group, and the user only needs to input the required voltage or capacity. For the user, the operation is very convenient. Under the same circuit, it can more conveniently meet the user's more test requirements.

Description

Technical Field

[0001] The present invention belongs to the technical field of power electronics, and particularly relates to a pulse capacitor series-parallel switching system based on the Monte Carlo method. Background Art

[0002] Pulse capacitors can charge and discharge quickly within a short time, releasing high-intensity pulse currents. Since pulse capacitors can provide instantaneous large current pulses and can simulate the pulse current situations that relays may encounter during actual operation, they are widely used in relay testing, especially in the testing of vehicle relays. For example, to test the arc resistance ability of relay contacts, the pulse current generated by the rapid discharge of a pulse capacitor is used to observe the performance of the contacts under such an impact and to test their lifespan and reliability. In addition, pulse capacitors can also be used to test the response time of relays. By generating pulses through capacitor discharge, it can be tested whether the closing or opening of the relay meets the expected fast response requirements, and so on.

[0003] However, in the relay test scenario, in order to conduct tests for various scenarios and test different relays, pulse capacitors with different output voltages and capacitance values are required. In traditional test devices, usually, users manually switch the series-parallel relationship of pulse capacitors to adjust in order to change the output or capacitance value, which has problems such as cumbersome operation, high difficulty, and easy errors. Although there are also operation guides, usually, the guides only cover several common voltages and capacitances and are difficult to meet the diverse test requirements of users. Summary of the Invention

[0004] An object of the present invention is to provide a pulse capacitor series-parallel switching system based on the Monte Carlo method for the above problems;

[0005] Another object of the present invention is to provide a pulse capacitor series-parallel switching method based on the Monte Carlo method.

[0006] To achieve the above object, the present invention adopts the following technical solutions:

[0007] A pulse capacitor series-parallel switching system based on the Monte Carlo method includes a pulse combination circuit, and the pulse combination circuit includes a plurality of pulse circuits directly or indirectly connected to each other;

[0008] Each pulse circuit includes a pulse capacitor and a power switch connected in series;

[0009] Each power switch is controlled by a control signal to conduct or turn off to change the connection mode of each pulse circuit in the pulse combination circuit;

[0010] The control signal is generated by the controller through executing the Monte Carlo search algorithm according to the required output voltage and / or total capacitance value.

[0011] In the above-mentioned pulse capacitor series-parallel switching system based on the Monte Carlo method, each pulse circuit includes a pulse capacitor, a power switch, and a resistor connected in series.

[0012] In the above-mentioned pulse capacitor series-parallel switching system based on the Monte Carlo method, the pulse capacitors of each pulse circuit have the same or different rated capacitance values;

[0013] The resistors of each pulse circuit have the same or different resistance values;

[0014] The pulse capacitors of each pulse circuit have the same or different rated voltages.

[0015] In the above-mentioned pulse capacitor series-parallel switching system based on the Monte Carlo method, the pulse combination circuit has a two-dimensional rectangular network topology;

[0016] Each power switch is controlled by a control signal to conduct or turn off, so as to control the corresponding pulse circuit to be cut off from the circuit, or to be separately switched into the circuit, or to be connected in series or in parallel with the remaining pulse circuits.

[0017] In the above-mentioned pulse capacitor series-parallel switching system based on the Monte Carlo method, the pulse combination circuit includes a number of parallel pulse circuits, and each parallel pulse circuit is formed by connecting a plurality of the above-mentioned pulse circuits in parallel;

[0018] Each parallel pulse circuit is connected in series in turn to form the two-dimensional rectangular network topology;

[0019] By controlling the power switches of each pulse circuit, the parallel number and parallel object of the pulse circuits of each parallel pulse circuit are controlled. If each pulse circuit has the same rated voltage, rated capacitance, and resistance value, then the parallel object doesn't matter.

[0020] In the above-mentioned pulse capacitor series-parallel switching system based on the Monte Carlo method, each parallel pulse circuit is also connected in parallel with a short-circuit switch, and the short-circuit switch is controlled by the control signal to conduct or turn off;

[0021] When the short-circuit switch is controlled to conduct, the corresponding parallel pulse circuit is disconnected. When the short-circuit switch is controlled to turn off, the parallel number and parallel object of the pulse circuits of the corresponding parallel pulse circuit are controlled by controlling the power switches of each pulse circuit.

[0022] A pulse capacitor series-parallel switching method for the above-mentioned pulse capacitor series-parallel switching system based on the Monte Carlo method, the method includes:

[0023] Obtain the current input current, as well as the rated capacitance value and rated voltage of each pulse circuit;

[0024] Generate a switch state matrix for the pulse combination circuit according to the circuit network of the pulse combination circuit;

[0025] Execute the Monte Carlo search algorithm according to the circuit network of the pulse combination circuit, input current, rated capacitance value, rated voltage, resistance value of each pulse circuit, and the required output voltage and / or total capacitance capacity to generate a policy tree and update the switch state matrix. The termination state of each path of the policy tree is that the circuit network is conducting and cannot be further expanded, and each path splits a switch state matrix;

[0026] Select the switch state matrix of the optimal policy according to the policy tree;

[0027] Generate the control signal according to the selected switch state matrix.

[0028] In the above pulse capacitance series-parallel switching method based on the Monte Carlo method, each element in the switch state matrix corresponds to a switch. When the element is in one state, it means the corresponding switch is in the conducting state, and when the element is in another state, it means the corresponding switch is in the off state;

[0029] The switch matrix contains the state information of each power switch, and when the parallel pulse circuit is also paralleled with a short-circuit switch, the switch matrix also contains the state information of the short-circuit switch.

[0030] In the above pulse capacitance series-parallel switching method based on the Monte Carlo method, the Monte Carlo search algorithm generates a policy tree and selects the optimal policy specifically including:

[0031] Use the UCB method to perform search and selection to obtain the child node with the maximum UCB value, and search downward like this until reaching the leaf node at the bottom of the policy tree;

[0032] Expand the leaf nodes that have not reached the termination state, and expand one or more nodes, which is the expansion operation;

[0033] Simulate the on-off states of each switch according to the random policy until the circuit is conducting, and generate a reward result. This is the simulation operation. After the expansion is completed, the simulation is carried out to simulate various conduction paths;

[0034] According to the simulation results, update the reward values of all nodes from bottom to top in reverse;

[0035] Iterate the above process to update the policy tree until reaching the specified time or the upper limit of computing power, and select the tree path with the highest reward value as the optimal policy according to the policy tree.

[0036] In the above pulse capacitance series-parallel switching method based on the Monte Carlo method, this method also includes establishing a correspondence table between the input current, total capacitance capacity, output voltage and the control signal;

[0037] When there are requirements for the same input current, the same total capacitance capacity, and output voltage, extract the corresponding control signal from the corresponding relationship table;

[0038] Otherwise, execute the Monte Carlo search algorithm to generate the control signal, and after the execution ends, add the corresponding relationship between the current input current, the obtained total capacitance capacity, output voltage, and the control signal to the corresponding relationship table.

[0039] The advantages of the present invention are as follows: The Monte Carlo method is adopted to realize the flexible control of the capacitor network, solving the technical problem of frequently changing the capacitance capacity in the vehicle relay test, and can realize the arbitrary switching of the series and parallel connections of the pulse capacitor bank to reach the target voltage / capacity gear.

[0040] The user only needs to input the required voltage or capacity. For the user, the operation is very convenient, and under the same circuit, it can more conveniently meet the user's more test requirements. Brief Description of the Drawings

[0041] Figure 1 is the system circuit diagram of an embodiment of the pulse capacitor series-parallel switching system based on the Monte Carlo method in Embodiment 1 of the present invention;

[0042] Figure 2 is the system structure block diagram of the pulse capacitor series-parallel switching system based on the Monte Carlo method in Embodiment 1 of the present invention;

[0043] Figure 3 is the schematic diagram of the strategy tree of the pulse capacitor series-parallel switching system based on the Monte Carlo method in Embodiment 1 of the present invention;

[0044] Figure 4 is the system circuit diagram of the pulse capacitor series-parallel switching system based on the Monte Carlo method of the present invention taking the pulse combination circuit composed of four parallel pulse circuits as an example;

[0045] Figure 5 is Figure 4 the schematic diagram of the strategy tree corresponding to the circuit;

[0046] Figure 6 is the system circuit diagram of the pulse capacitor series-parallel switching system based on the Monte Carlo method in Embodiment 2 of the present invention;

[0047] Figure 7 is the system circuit diagram of the pulse capacitor series-parallel switching system based on the Monte Carlo method in Embodiment 2 of the present invention taking the pulse combination circuit composed of four parallel pulse circuits as an example;

[0048] Figure 8 is Figure 7 the schematic diagram of the strategy tree corresponding to the circuit.

[0049] Reference numerals: Pulse combination circuit 1, parallel pulse circuit 11, pulse circuit 2, controller 3. Detailed implementation

[0050] Embodiment 1

[0051] This solution provides a pulse capacitor series - parallel switching system based on the Monte Carlo method and provides a corresponding implementation method. As Figure 1 and Figure 2 shown, the system includes a pulse combination circuit 1. The pulse combination circuit 1 includes a number of parallel pulse circuits 11. Each parallel pulse circuit 11 is formed by connecting multiple pulse circuits 2 in parallel; each parallel pulse circuit 11 is connected in series in sequence to form a two - dimensional rectangular network topology. Each pulse circuit 2 includes a pulse capacitor, a power switch, and a resistor connected in series. In this embodiment, the power switch uses a relay, and when in use, it is not limited to a relay.

[0052] The conduction and cutoff of the relay are controlled by a control signal, thereby changing the connection mode of the pulse capacitors, and outputting the required capacitor voltage and capacitance. Here, the capacitor voltage and capacitance output are for the entire pulse combination circuit 1, that is, the output voltage and total capacitance of this combination circuit. The control signal of the relay is generated by the controller 3 using the Monte Carlo search algorithm. According to Figure 1 the states of each relay in the circuit network shown, the state matrix V is obtained as follows:

[0053]

[0054] For different circuit networks, there will be different state matrices. The state matrix contains the states of each power switch in the circuit network. In the above formula, each element v ij value corresponds to the state of a relay. Taking "1" means the relay is conducting, and taking "0" means the relay is off.

[0055] The pulse capacitors of each pulse circuit 2 have the same or different rated capacitance values as required. The resistors of each pulse circuit 2 have the same or different resistance values as required. The pulse capacitors of each pulse circuit 2 have the same or different rated voltages as required. The greater the number of different rated capacitances, resistance values, and rated voltages, the more output voltage and capacitance capacity levels can be achieved by the same circuit network, and more diverse test requirements can be met.

[0056] By controlling the power switches of each pulse circuit 2, the number of pulse circuits 2 connected in parallel and the parallel objects of each parallel pulse circuit 11 are controlled to output the required voltage and capacitance.

[0057] As Figure 3As shown in the figure, the method for generating a control signal using the Monte Carlo search algorithm in this solution is as follows:

[0058] During the execution of the algorithm, a policy tree is generated. The policy tree has a root node and child nodes. The termination state of each path in the policy tree is determined by whether the corresponding policy tree can continue to expand and whether the circuit network has been completely conducted. Conductance does not necessarily mean termination; it may still be possible to continue expanding, while termination must mean conductance. For different paths, the path lengths can be the same or different. Eventually, there are as many available switch state matrices as there are conductive paths, and the one with the highest score is selected from these switch state matrices. The Monte Carlo search algorithm, based on the capacitance state within the system, completes the selection algorithm, expansion algorithm, simulation algorithm, and backpropagation algorithm through the policy tree, and finally selects the capacitance access strategy according to the node reward value to generate the corresponding control signal.

[0059] The Monte Carlo search algorithm uses the Upper Confidence Bounds (UCB) method for searching, which is given by the following formula:

[0060]

[0061] Among them, Q is the reward value of the node, N is the number of times the parent node has been visited, n represents the number of times the current node has been visited, and c is an adjustable parameter. In the selection algorithm, starting from the root node, the child node with the maximum value calculated using UCB is selected and searched downward in this way until reaching the leaf node at the bottom of the tree, waiting for the next operation. After reaching the leaf node, if the termination state has not been reached (the termination state means that it is judged through the state matrix V that the pulse circuit network has been completely conducted and cannot continue to expand), then this node is expanded to generate one or more nodes. Then, based on the current state, the subsequent capacitance access is simulated according to the random policy until the circuit is conducted and the reward value result is generated. The reward value is given by the following formula:

[0062]

[0063] In the formula, a is the obtained voltage, b is the obtained capacitance, a 0 is the required voltage, b 0 is the required capacitance. a, bIt is calculated based on the circuit network, i.e., the connection relationship of each pulse circuit 2, as well as the current input current, the rated capacitance value, rated voltage, resistance value, etc. of each pulse circuit 2 connected to the current state. This calculation is a conventional algorithm and will not be elaborated here. As can be seen from the formula, the reward value refers to the degree of closeness between the obtained voltage and capacitance and the required voltage and capacitance. The closer they are, the greater the reward value, and vice versa. Here, taking the simultaneous setting of the required voltage and capacitance as an example, when only the voltage or capacitance is set, the corresponding parameters are used to calculate the reward value. For example, if an example of a required output voltage of 10V is given later and the capacitance is not specifically set, at this time, the formula is:

[0064]

[0065] According to the simulation results, from bottom to top, the reward values of all nodes are updated in reverse.

[0066] When the specified time or the upper limit of computing power is reached, the algorithm stops, selects the node with the largest reward value to execute and ends the algorithm, and outputs the state information V at this time 1 . According to the state, the access strategy selection of the capacitor is completed, and the corresponding control signal is generated.

[0067] This solution proposes a pulse capacitor series-parallel switching system based on the Monte Carlo method. The basic principle of this solution is as follows: The Monte Carlo method uses exploration and exploitation to solve the problem of quickly obtaining the required voltage and capacitance in the pulse capacitor series-parallel switching system. By exploring, moving forward in an unknown direction, and continuously trying new circuit connection methods to continuously obtain better strategies. At the same time, using the experience obtained from the previous attempts, a reliable capacitor access strategy selection is obtained, and the final optimal capacitor access strategy is obtained through multiple simulations and attempts. The user only needs to set the required output voltage and / or capacitance.

[0068] The following is an example for illustration using a pulse combination circuit 1 composed of four parallel pulse circuits 11:

[0069] As Figure 4 shown, in this example, the rated voltages of the four parallel pulse circuits 11 are respectively , the pulse capacitor capacitances are all 1C, and the user needs to achieve a 10V voltage output and obtain the maximum capacitance. The initial state matrix of the capacitor network can be

[0070]

[0071] The generated strategy tree is as Figure 5 shown. After the selection algorithm, expansion algorithm, simulation algorithm, and backpropagation algorithm, finally, when the computing power upper limit is reached, the algorithm stops, selects the node with the largest reward value to execute, and outputs the state information V at this time 1 is

[0072]

[0073] Reach a capacity of 2 / 3C.

[0074] According to the state matrix V 1 The states of each switch can be obtained, and control signals are output to each switch respectively to complete the series-parallel switching of the pulse capacitors.

[0075] This solution realizes the series-parallel switching of pulse capacitors through the Monte Carlo method, can realize arbitrary series-parallel switching of pulse capacitor banks, flexibly and reliably output the required voltage and capacity, and solve the disadvantages of traditional capacitor banks such as the need for frequent manual adjustment or limited test capabilities.

[0076] Embodiment 2

[0077] As Figure 6 shown, this embodiment is similar to Embodiment 1, except that in each parallel pulse circuit 11 of this embodiment, a short-circuit switch is further connected in parallel, that is Figure 6 S17, S27, S37 in, and the rest of S11-S16, S21-S26, S31-S36 are all the above-mentioned power switches, and the short-circuit switch is also controlled to conduct or turn off by a control signal, and the switch matrix contains the state information of each power switch and short-circuit switch.

[0078] When the short-circuit switch is controlled to conduct, the corresponding parallel pulse circuit 11 is disconnected. When the short-circuit switch is controlled to turn off, the number of parallel pulse circuits 2 and the parallel objects of the pulse circuit 2 in the corresponding parallel pulse circuit 11 are controlled by controlling the power switches of each pulse circuit 2.

[0079] The following also takes the pulse combination circuit 1 composed of four parallel pulse circuits 11 as an example for illustration:

[0080] As Figure 7 shown, in this example, the rated voltages of the four parallel pulse circuits 11 are respectively , and the user needs to realize a capacitance capacity of 2C. The initial state matrix of the capacitance network can be

[0081]

[0082] The generated policy tree is as Figure 8 shown. After passing through the selection algorithm, expansion algorithm, simulation algorithm, and backpropagation algorithm, finally when the calculation capacity limit is reached, the algorithm stops, and the node with the largest reward value is selected for execution, and the state information V at this time is output 1 is

[0083]

[0084] Reach a capacity of 2C and a voltage of 5V.

[0085] According to the state matrix V 1 The states of each switch can be obtained, and control signals are output to each switch respectively to complete the series - parallel switching of the pulsed capacitors.

[0086] Embodiment III

[0087] This embodiment is similar to Embodiment I, except that the series - parallel control method of this embodiment further includes:

[0088] Establish a correspondence table of input current, total capacitance, output voltage and control signal as shown in Table 1 below;

[0089] When there are the same input current, and the same requirements for total capacitance and output voltage, extract the corresponding control signal from the said correspondence table;

[0090] Otherwise, execute the Monte Carlo search algorithm to generate a control signal. The control signal preferably conducts at the highest voltage. And after the execution is completed, add the correspondence of the current input current and the obtained total capacitance, output voltage and control signal to the said correspondence table.

[0091] Table 1 Correspondence Table

[0092]

[0093] That is to say, during the user's use, a pre - storage table belonging to the customer is generated and improved according to the customer's needs. When the user needs the same output voltage or capacitance, the control signal can be quickly extracted without calculation and optimal strategy selection. At the same time, when the user has a new test requirement for output voltage or capacitance, the Monte Carlo search algorithm is executed to automatically calculate and generate a control signal, and this control signal is used to further improve the pre - storage table. In this way, both speed and diversification can be taken into account, and the potential of the output voltage and output capacity of the pulsed capacitor series - parallel switching system can be deeply explored.

[0094] The specific embodiments described herein are merely illustrative of the spirit of the present invention. Those skilled in the art to which the present invention pertains can make various modifications or supplements to the described specific embodiments or use similar ways to replace them, but will not deviate from the spirit of the present invention or exceed the scope defined by the appended claims.

[0095] Although terms such as pulsed combination circuit 1, parallel pulsed circuit 11, pulsed circuit 2, controller 3, etc. are used more frequently in this article, the possibility of using other terms is not excluded. Using these terms is only for more convenient description and explanation of the essence of the present invention; interpreting them as any additional limitation is contrary to the spirit of the present invention.

Claims

1. A pulse capacitor series-parallel switching method based on Monte Carlo method, characterized in that: The method is implemented by a pulse capacitor series-parallel switching system based on the Monte Carlo method, wherein the pulse capacitor series-parallel switching system comprises a pulse combination circuit (1), and the pulse combination circuit (1) comprises a plurality of pulse circuits (2) directly or indirectly connected to each other; Each pulse circuit (2) comprises a pulse capacitor and a power switch connected in series with each other; Each power switch is turned on or off by a control signal to change the connection mode of each pulse circuit (2) in the pulse combination circuit (1); The control signal is generated by the controller (3) by executing a Monte Carlo search algorithm according to the required output voltage and / or total capacitance; The pulse capacitor series-parallel switching method includes: Obtain the current input current, as well as the rated capacitance and rated voltage of each pulse circuit (2); generating a switch state matrix for the pulse combination circuit (1) according to the circuit network of the pulse combination circuit (1); According to the circuit network of the pulse combination circuit (1), the input current, the rated capacitance value, the rated voltage, the resistance value of each pulse circuit (2), and the required output voltage and / or the total capacitance, a Monte Carlo search algorithm is executed to generate a strategy tree and update the switch state matrix, wherein the termination state of each path of the strategy tree is that the circuit network is turned on and cannot be further expanded, and each path splits a switch state matrix; Select the switch state matrix of the optimal strategy according to the strategy tree; The control signal is generated according to the selected switch state matrix.

2. The pulse capacitor series-parallel switching method based on the Monte Carlo method according to claim 1, characterized in that: Each element in the switch state matrix corresponds to a switch. When the element is in one state, it indicates that the corresponding switch is in the on state. When the element is in another state, it indicates that the corresponding switch is in the off state. The switch matrix contains status information of each power switch, and when the parallel pulse circuit (11) is also connected in parallel with a short-circuit switch, the switch matrix also contains status information of the short-circuit switch.

3. The pulse capacitor series-parallel switching method based on the Monte Carlo method according to claim 1, characterized in that: The Monte Carlo search algorithm generates a strategy tree and selects the optimal strategy, including: Use the UCB method to search and select the child node with the maximum UCB value, and search downward until you reach the leaf node at the bottom of the strategy tree; Expand the leaf nodes that have not reached the terminal state to expand one or more nodes; According to the random strategy, simulate the on and off states of each switch until the circuit is turned on and a reward result is generated; According to the simulation results, update the reward values ​​of all nodes in reverse order from bottom to top; The above process is iterated to update the strategy tree until the specified time or computing power limit is reached, and the tree path with the highest reward value is selected as the optimal strategy according to the strategy tree.

4. The pulse capacitor series-parallel switching method based on the Monte Carlo method according to claim 3 is characterized in that: The method further includes establishing a corresponding relationship table of input current, total capacitance, output voltage and control signal; When the same input current, total capacitance and output voltage are required, the corresponding control signal is extracted from the corresponding relationship table; Otherwise, the Monte Carlo search algorithm is executed to generate a control signal, and after the execution is completed, the corresponding relationship between the input current and the obtained total capacitance, output voltage and control signal is added to the corresponding relationship table.

5. The pulse capacitor series-parallel switching method based on the Monte Carlo method according to claim 1, characterized in that: Each pulse circuit (2) comprises a pulse capacitor, a power switch and a resistor which are connected in series.

6. The method for switching pulse capacitors in series and in parallel based on the Monte Carlo method according to claim 5, characterized in that: The pulse capacitors of the pulse circuits (2) have the same or different rated capacitance values; The resistors of each pulse circuit (2) have the same or different resistance values; The pulse capacitors of the pulse circuits (2) have the same or different rated voltages.

7. The pulse capacitor series-parallel switching method based on the Monte Carlo method according to claim 6, characterized in that: The pulse combination circuit (1) presents a two-dimensional rectangular network topology structure; Each power switch is turned on or off by a control signal to control the corresponding pulse circuit (2) to be cut off from the circuit, or to be cut into the circuit alone, or to be connected in series or in parallel with other pulse circuits.

8. The pulse capacitor series-parallel switching method based on the Monte Carlo method according to claim 7, characterized in that: The pulse combination circuit (1) comprises a plurality of parallel pulse circuits (11), each of which is formed by connecting a plurality of the pulse circuits (2) in parallel; The parallel pulse circuits (11) are sequentially connected in series to form the two-dimensional rectangular network topology structure; The parallel quantity and parallel object of the pulse circuits (2) of each parallel pulse circuit (11) are controlled by controlling the power switch of each pulse circuit (2).

9. The method for switching pulse capacitors in series and in parallel based on the Monte Carlo method according to claim 8, characterized in that: Each parallel pulse circuit (11) is also connected in parallel with a short-circuit switch, and the short-circuit switch is turned on or off by the control signal; When the short-circuit switch is controlled to be turned on, the corresponding parallel pulse circuit (11) is disconnected; when the short-circuit switch is controlled to be turned off, the parallel number and parallel object of the pulse circuit (2) of the corresponding parallel pulse circuit (11) are controlled by controlling the power switch of each pulse circuit (2).

Images