Aluminum electrolytic cell series ground leakage fault location, early warning detection system and method

Abstract

The present invention discloses a system and method for locating and early warning of earth leakage faults in an aluminum electrolytic cell series. Each aluminum electrolytic cell is connected with an earth leakage fault detection circuit, and the earth leakage fault detection circuit includes a signal transmission wire, a voltage receiving wire, an insulation resistor and a main control machine; one end of the signal transmission wire is connected to the main control machine, and the other end of each signal transmission wire is connected to the grounding point of the corresponding aluminum electrolytic cell; the grounding point is also connected to the corresponding aluminum electrolytic cell through an insulation resistor; the grounding point is also connected to one end of the voltage receiving wire, and the other end of the voltage receiving wire is connected to the main control machine, the main control machine calculates the resistance value of the insulation resistor, and determines whether the corresponding aluminum electrolytic cell has an earth leakage fault, and accurately locates the aluminum electrolytic cell with an earth leakage fault; at the same time, an LSTM network is used to predict the changes in the insulation resistance of the aluminum electrolytic cell and each point to realize an early warning of an earth leakage fault. The present invention can timely and accurately locate and early warning the earth leakage faults of the aluminum electrolytic cell and each point.

Description

Technical Field

[0001] The present invention relates to the technical field of ground leakage fault detection for aluminum electrolytic cells, and in particular to a ground leakage fault positioning and early warning detection system and method for an aluminum electrolytic cell series. Background Art

[0002] When aluminum electrolysis enterprises produce electrolytic aluminum, the aluminum electrolytic cells of the aluminum electrolytic plant work by supplying power to multiple aluminum electrolytic cells connected in series through a DC high-voltage busbar (voltage ranges from several hundred volts to thousands of volts), using DC power to electrolyze aluminum and generate aluminum liquid for output. The power supply voltage of each aluminum electrolytic cell is about 4V, but the power supply current is very large, ranging from tens of thousands of amperes to hundreds of thousands of amperes, so the entire DC high-voltage busbar passage is equivalent to a pure short-circuited conductor. In order to ensure the safe operation of the aluminum electrolytic cell, the aluminum electrolytic cell is supported by an insulator component between the aluminum electrolytic cell and the ground. With the long-term operation of the electrolytic aluminum, the supporting insulator will gradually deteriorate in performance, resulting in poor insulation of the aluminum electrolytic cell to the ground, and the insulation resistance will become smaller. In severe cases, it will cause leakage of electricity from the aluminum electrolytic cell to the ground. Electricity leakage from any point of the aluminum electrolysis cell to the ground may cause electric shock to the operator or damage to the electrical equipment related to the aluminum electrolysis cell. If these faults are not discovered and eliminated in time, the aluminum electrolysis cell may even be scrapped. At the same time, electricity leakage from the aluminum electrolysis cell to the ground causes electrical energy to flow into the earth, causing energy waste, resulting in significant economic losses and serious safety accidents to the enterprise.

[0003] At present, the judgment of the insulation fault of aluminum electrolytic cell series to the ground at home and abroad mostly adopts the busbar midpoint voltage drift detection method and the single-frequency signal injection detection method. However, the midpoint voltage drift detection method can only infer the approximate location of the aluminum electrolytic cell where the aluminum electrolytic cell leakage accident occurred, and it is difficult to accurately locate a certain aluminum electrolytic cell, while the single-frequency signal injection detection method can only detect the occurrence of the aluminum electrolytic cell series leakage fault, and cannot accurately locate the specific fault slot and fault point. Both methods cannot detect the resistance value of the insulation resistance of each slot in the aluminum electrolytic cell series and each point of each electrolytic cell to the ground in real time online. At the same time, since the resistance value of the insulation resistance of each slot and each point to the ground cannot be detected online in real time, the resistance value change of the insulation resistance of each slot in the aluminum electrolytic cell series and each point of each electrolytic cell to the ground cannot be monitored online in real time, and the accurate positioning of the leakage fault of each slot in the aluminum electrolytic cell series and each point of each electrolytic cell cannot be realized, and the early warning detection of the leakage fault of each slot in the aluminum electrolytic cell series and each point of each electrolytic cell cannot be realized.

[0004] For example, the patent with the authorization publication number CN1948975B in 2011, entitled "A device and method for detecting the insulation resistance of an electrolytic cell series to the ground", and the patent application with the application publication number CN112034283A in 2020, entitled "Device and system and process for detecting and locating grounding faults in aluminum electrolytic cells". Both of the above patents belong to the single-frequency signal injection detection method, which can only detect the occurrence of ground leakage faults in the aluminum electrolytic cell series, but cannot accurately locate the specific faulty cell and fault point, nor can it detect the insulation resistance value of each electrolytic cell and each point of each electrolytic cell online in real time, and cannot realize the early warning detection of ground leakage faults in each cell of the aluminum electrolytic cell series and each point of each electrolytic cell.

[0005] In summary, the currently reported technologies and methods cannot measure the insulation resistance value of the aluminum electrolytic cell to the ground, so it is impossible to know the insulation performance of the electrolytic cell and accurately determine and locate the leakage fault point of the aluminum electrolytic cell to the ground, and it is even more impossible to predict the performance change trend of the deterioration of the insulation performance of the aluminum electrolytic cell. Summary of the invention

[0006] In view of the problem that the prior art cannot accurately locate the ground leakage fault point of an aluminum electrolysis cell series, the present invention proposes a ground leakage fault positioning, early warning detection system and method for an aluminum electrolysis cell series. The aluminum electrolysis cells are numbered, and ground leakage fault detection circuits are respectively set to measure the resistance value of the aluminum electrolysis cell's insulation resistance to the ground, and whether the corresponding aluminum electrolysis cell has a ground leakage fault is determined according to the resistance value, thereby accurately locating the ground leakage fault point of the aluminum electrolysis cell.

[0007] In order to achieve the above object, the present invention provides the following technical solutions:

[0008] A ground leakage fault location and early warning detection system for an aluminum electrolysis cell series includes a voltage source, a DC bus, and i aluminum electrolysis cells connected in series through the DC bus, characterized in that each of the aluminum electrolysis cells is connected to a ground leakage fault detection circuit, and the ground leakage fault detection circuit includes a signal sending wire, a voltage receiving wire, an insulation resistor and a main control computer processing unit;

[0009] One end of the signal sending wire is connected to the main control computer processing unit, and the other end of each signal sending wire is connected to the grounding point of the corresponding aluminum electrolytic cell; the grounding point is also connected to the corresponding aluminum electrolytic cell through an insulation resistor; the grounding point is also connected to one end of the voltage receiving wire, and the other end of the voltage receiving wire is connected to the main control computer processing unit. The main control computer processing unit calculates the resistance value of the insulation resistor, thereby locating the aluminum electrolytic cell where a ground leakage fault occurs.

[0010] Preferably, each aluminum electrolysis cell has a unique number, and each number corresponds to a unique installation location.

[0011] Preferably, the signal sending wire is connected to the main control computer processing unit through a current limiting resistor.

[0012] Preferably, the main control computer processing unit includes a signal generating module, a sending module, a receiving module, a calculating module, a judging module, a predicting alarm module and a controller, and the signal generating module, the sending module, the receiving module, the calculating module, the judging module and the predicting alarm module are connected to the controller respectively;

[0013] A signal generating module is used to generate a measurement signal; a sending module is used to send the measurement signal to the grounding point of the aluminum electrolytic cell to be tested; a receiving module is used to receive the voltage of the grounding point; a calculating module is used to calculate the resistance value of the insulation resistance corresponding to the aluminum electrolytic cell to be tested according to the voltage of the grounding point; a judging module is used to compare the resistance value calculated by the calculating module; and a predictive alarm module is used to issue an alarm prompt according to the result of the comparing module.

[0014] Preferably, the measurement signal generated by the signal generating module is a sinusoidal multi-carrier FSK signal, and the frequency range of the measurement signal is 500 Hz to 5 KHz.

[0015] Preferably, the main control computer processing unit also includes an early warning module for predicting changes in insulation resistance based on the LSTM neural network and historical insulation resistance values, thereby realizing an early warning of ground leakage faults in the aluminum electrolysis cell.

[0016] The present invention also provides a method for locating and early warning of earth leakage faults in an aluminum reduction cell series, which is characterized by comprising the system for locating and early warning of earth leakage faults in an aluminum reduction cell series according to any one of claims 1 to 6, and specifically comprising the following steps:

[0017] S1: When testing the aluminum electrolytic cell Ci-1 to be tested, the main control computer processing unit generates a measurement signal and sends it to the grounding point i-1 through the i-1th signal sending wire, and generates a voltage Vi-1 at the grounding point i-1;

[0018] S2: The measurement signal returns to the main control computer processing unit after passing through the i-1th insulation resistor and the aluminum electrolytic cell Ci-1 to be tested; the i-1th voltage receiving wire feeds back the voltage Vi-1 to the main control computer processing unit, and the main control computer processing unit simultaneously receives the voltage of the grounding point of other aluminum electrolytic cells;

[0019] S3: The main control computer processing unit calculates the resistance value of the i-1th insulation resistor according to the received voltage and compares it with the threshold value. If the insulation resistance value is less than the threshold value, it means that the aluminum reduction cell has a fault and an alarm should be given immediately. At the same time, the number of the aluminum reduction cell is displayed to locate the aluminum reduction cell with a ground leakage fault.

[0020] Preferably, the resistance value of the (i-1)th insulation resistor is calculated as:

[0021] S3-1: Calculate the resistance RG between the grounding point i-1 of the i-1th insulation resistance and the adjacent grounding point i-2 a And the resistance RG between ground point i-1 and the adjacent ground point i b :

[0022]

[0023] In formula (1), RG a Represents the resistance between ground point i-1 and ground point i-2; RG b Represents the resistance between ground point i-1 and ground point i; V i-1 Represents the voltage of ground point i-1; V i-2 Represents the voltage of ground point i-2; V in Indicates the measured signal voltage; V i represents the voltage at the grounding point i; R0 represents the resistance value of the current limiting resistor;

[0024] S3-2: Calculate the resistance value Z of the insulation resistance Zi-2 and Zi adjacent to the i-1 insulation resistance i-2 , Z i :

[0025]

[0026] In formula (2), Z i-2 represents the resistance value of the i-2nd insulation resistor Zi-2 adjacent to the i-1th insulation resistor, Z i represents the resistance value of the i-th insulation resistor Zi adjacent to the i-1-th insulation resistor;

[0027] S3-3: Calculate the resistance value Z of the i-1th insulation resistor Zi-1 i-1 :

[0028] Calculate first Ra=RG a +Z i-2 , Rb=RG b +Z i ,

[0029] Then let R ab =R a / / R b , because R z =R ab / / Z i-1 ,

[0030] Then we can get:

[0031] Solve to get the resistance value Z of the i-1th insulation resistor Zi-1i-1 :

[0032] In formula (3), / / represents the resistance parallel operator, R z The total equivalent resistance Ra and Rb of the i-1th grounding point are intermediate variables introduced for the convenience of calculation.

[0033] Preferably, it includes S4: using an LSTM neural network to calculate the historical insulation resistance value of each aluminum electrolytic cell collected by the main control computer processing unit, and outputting a predicted value of the insulation resistance.

[0034] Preferably, the calculation of the predicted value of the insulation resistance comprises the following steps:

[0035] S4-1: Construct an LSTM-based aluminum electrolysis cell insulation resistance prediction network;

[0036] S4-2: Obtain the historical insulation resistance values ​​of the aluminum electrolytic cell stored in the main control computer processing unit and divide them into a training set, a validation set and a test set;

[0037] S4-3: Initialize the parameters of the LSTM-based aluminum electrolysis cell insulation resistance prediction network model:

[0038] The network output y and the initial value c <t-1>< / t-1> Initialize to 0; the weight matrix w of the network hidden layer c and bias b c , the weight matrix w of the input gate i and bias b i , update the gate weight matrix w u and bias b u , the weight matrix w of the output gate o and bias b o And the weight matrix w of the forget gate f and bias b f Initialize to a random number between 0 and 1; set the number of neurons in the network input layer to i+1, and the number of neurons in the network hidden layer, input gate, forget gate, and output gate to i; c <t>< / t> Represents the memory information at time t; a <t>< / t> represents the output of the model at time t; x <t>< / t> Represents the input of the model at time t;

[0039] S4-4: Training of the LSTM-based aluminum electrolysis cell insulation resistance prediction network model:

[0040] The model is trained using the divided sample training set, and the model weights and biases are optimized using the reverse tuning algorithm J according to the difference between the predicted value and the true value;

[0041] In a single LSTM unit structure, the forget gate Γ is used f <t>< / t> According to the output a at time t-1 <t-1>< / t-1> and the input x at time t <t>< / t> Calculate the amount of information inherited from the previous time step:

[0042] Γ f <t>< / t> =σ(W f [a <t-1>< / t-1> ,x <t>< / t> ]+b f );

[0043] At the same time according to a <t-1>< / t-1> and x <t>< / t> Calculate the memory information that needs to be updated Update gate Γ u <t>< / t> calculate The proportion of node memory information, the final node memory information c <t>< / t> Determined by the forget gate and the update gate:

[0044]

[0045] Γ u <t>< / t> =σ(W u [a <t-1>< / t-1> ,x <t>< / t> ]+b u ),

[0046]

[0047] Then the model output is determined by the output gate Γ o <t>< / t> Calculated from node memory information

[0048] Γ o <t>< / t> =σ(W o [a <t-1>< / t-1> ,x <t>< / t> ]+b o ), a <t>< / t> =Γ o <t>< / t> ·tanh(c <t>< / t> ), σ represents the sigmoid function;

[0049] S4-5: Input the collected insulation resistance value of the aluminum electrolytic cell into the trained model to obtain a predicted value of the insulation resistance.

[0050] In summary, due to the adoption of the above technical solution, compared with the prior art, the present invention has at least the following beneficial effects:

[0051] The present invention can monitor the insulation resistance value of the aluminum electrolytic cell series to the ground in real time online, and can judge the insulation performance of the aluminum electrolytic cell according to the insulation resistance value, whether insulation failure will occur, and accurately locate the aluminum electrolytic cell with ground leakage fault, so as to inform the maintenance personnel to eliminate the fault in time, avoid accidents, ensure the safety of operators and the efficient and normal operation of the aluminum electrolytic cell. At the same time, the resistance value detected by positioning is used to predict the changes in the insulation resistance of the aluminum electrolytic cell and each point of the aluminum electrolytic cell to the ground through the LSTM neural network, so as to realize the early warning of ground leakage fault. The present invention can timely and accurately detect, locate and warn the ground leakage fault of a single aluminum electrolytic cell or multiple aluminum electrolytic cells in the aluminum electrolytic cell series and each point of the aluminum electrolytic cell; improve the current efficiency, maintenance efficiency and energy utilization rate, and reduce the production and maintenance costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 The present invention is a schematic diagram of a system for locating and early warning of ground leakage faults in a series of aluminum electrolysis cells according to an exemplary embodiment of the present invention.

[0053] Figure 2 It is a schematic diagram of a method for locating and detecting ground leakage faults in a series of aluminum electrolysis cells according to an exemplary embodiment of the present invention.

[0054] Figure 3 FIG. 4 is a schematic diagram of an insulation resistance equivalent circuit according to an exemplary embodiment of the present invention.

[0055] Figure 4 The present invention is a schematic diagram of a method for early warning of ground leakage faults in a series of aluminum electrolysis cells according to an exemplary embodiment of the present invention.

[0056] Figure 5 The figure is a schematic diagram of a network for predicting the insulation resistance of an aluminum electrolysis cell series to ground based on LSTM according to an exemplary embodiment of the present invention. DETAILED DESCRIPTION

[0057] The present invention is further described in detail below in conjunction with the examples and specific implementation methods. However, this should not be understood as the scope of the above subject matter of the present invention being limited to the following examples, and all technologies realized based on the content of the present invention belong to the scope of the present invention.

[0058] In the description of the present invention, it is necessary to understand that the terms "longitudinal", "lateral", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", etc., indicating the orientation or position relationship, are based on the orientation or position relationship shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention.

[0059] like Figure 1 As shown, the present invention provides a system for locating and detecting earth leakage faults of an aluminum reduction cell series, comprising a voltage source, a DC bus, and m aluminum reduction cells connected in series through the DC bus to form an aluminum reduction cell series (the aluminum reduction cells are numbered C1, C2, ..., C i ,C i+1 ,...,C m-2 ,C m-1 ,C m ), each aluminum electrolysis cell is connected to a ground leakage fault detection circuit.

[0060] The ground leakage fault detection circuit includes a signal transmission wire, a voltage receiving wire, an insulation resistor, a grounding resistor, a current limiting resistor and a main control computer processing unit U1 (which can be referred to as the main control computer). One end of the signal transmission wire is connected to the main control computer processing unit, and the other end of each signal transmission wire is connected to the grounding point of the corresponding aluminum electrolytic cell; the grounding point is also connected to the corresponding aluminum electrolytic cell through the insulation resistor; the grounding point is also connected to one end of the voltage receiving wire, and the other end of the voltage receiving wire is connected to the main control computer processing unit. The main control computer processing unit calculates the resistance value of the insulation resistor, thereby locating the aluminum electrolytic cell with the ground leakage fault and the specific fault point of the aluminum electrolytic cell (such as the damper, inlet and outlet, etc.); the grounding point is also grounded through the grounding resistor.

[0061] That is, the system includes m signal transmission wires (T1, T2, ..., T i ,T i+1 ,...,T m-2 ,T m-1 ,T m ),m voltage receiving wires (R1, R2, ..., R i ,R i+1 ,...,R m-2 ,R m-1 ,R m ), m insulation resistances (Z1, Z2, ..., Z i ,Z i+1 ,...,Z m-2 ,Z m-1 ,Z m), m grounding resistors (RG1, RG2, ..., RG i ,RG i+1 ,...,RG m-2 ,RG m-1 ,RG m ), m current limiting resistors (Ro1, Ro2, ..., Ro m ) and main control computer processing unit U1.

[0062] In this embodiment, each aluminum electrolytic cell has a unique number, and each number corresponds to a unique position. The signal transmission wire and the voltage receiving wire are both high temperature fluorine-proof twisted pair wires.

[0063] The mth signal transmission wire T among the m signal transmission wires m One end of the mth current limiting resistor RO m One end of the mth current limiting resistor RO m The other end is connected to the sending module of the main control computer processing unit; the mth signal sending wire T m The other end of the corresponding m-th aluminum electrolysis cell C m The grounding point m is also connected to the mth voltage receiving wire R m The mth voltage receiving wire R m The other end is connected to the receiving module of the main control computer processing unit. The grounding point m is also connected to the mth insulation resistance Z m One end and the mth grounding resistor RG m One end of the mth grounding resistor RG is connected m The other end of the mth insulation resistance Z m The other end is connected to the mth aluminum electrolysis cell C m connect.

[0064] For example, one end of the first signal transmission wire T1 is connected to one end of the first current limiting resistor Ro1, and the other end of the first current limiting resistor Ro1 is connected to the transmission module of the main control computer processing unit; the other end of the first signal transmission wire T1 is connected to the grounding point 1 of the corresponding first aluminum electrolytic cell C1, and the grounding point 1 is also connected to one end of the first voltage receiving wire R1, and the other end of the first voltage receiving wire R1 is connected to the receiving module of the main control computer processing unit. The grounding point 1 is also connected to one end of the first insulation resistor Z1 and one end of the first grounding resistor RG1 respectively, the other end of the first grounding resistor RG1 is grounded, and the other end of the first insulation resistor Z1 is connected to the first aluminum electrolytic cell C1.

[0065] The signal generation module of the main control computer processing unit generates a measurement signal, and the signal transmission module transmits the measurement signal to the grounding point of the corresponding aluminum electrolytic cell through the signal transmission wire. After the measurement signal passes through the insulation resistance, it passes through the aluminum electrolytic cell to be measured and the DC bus and then returns to the Vs end of the main control computer processing unit through the S-BUS. And the measurement signal will generate a voltage (V1, V2, ..., V i ,V i+1 ,...,V m-2 ,V m-1 ,V m ), the voltage will be returned to the main control computer processing unit through the voltage receiving wire. The main control computer processing unit calculates the insulation resistance value according to the voltage, and compares it with the threshold value required by the electrolytic aluminum industry for the insulation resistance of the electrolytic cell to the ground to determine whether the aluminum electrolytic cell has poor insulation or insulation damage. If the insulation resistance value is less than the threshold value, it means that the aluminum electrolytic cell will have a ground leakage fault and an alarm should be sounded immediately. At the same time, the number of the aluminum electrolytic cell will be displayed, thereby locating the aluminum electrolytic cell with a ground leakage fault; if the insulation resistance value is greater than or equal to the threshold value, it means that the aluminum electrolytic cell does not have a ground leakage fault.

[0066] The busbar current at the aluminum electrolysis site is very large, up to hundreds of thousands of amperes, so the electromagnetic interference it generates is very serious. To resist interference, the measurement signal sent uses a sinusoidal multi-carrier FSK signal format, generally using four carrier frequencies (such as 500Hz, 2.4KHz, 4KHz, 5KHz). Four frequencies can measure four resistance values. From the change of the resistance value with the frequency, it can be determined whether the insulation resistance is capacitive or inductive. According to its characteristics, it can be determined which frequency the resistance value measured is more suitable for the final insulation resistance value. At the same time, multiple frequency points are also beneficial for on-site measurement anti-interference. Those frequencies outside the interference frequency band can be selected as the measurement signal frequency. The frequency range is from 500Hz to 5KHz, and the specific frequency is determined according to the on-site interference spectrum diagram. The carrier change of the measurement signal is switched at a fixed periodic frequency (4Hz), and the initial phase of each carrier is the same, the signal amplitude is not greater than 15V, and the drive current is less than 1A.

[0067] Take the first aluminum electrolysis cell as an example:

[0068] The signal generation module of the main control computer processing unit generates a measurement signal, and the signal sending module sends the measurement signal to the grounding point 1 of the aluminum electrolytic cell C1 through the signal sending wire T1. The measurement signal passes through the insulation resistor Z1, passes through the aluminum electrolytic cell C1 and the DC bus, and then returns to the main control computer processing unit. At the same time, the measurement signal will generate a voltage V1 at the grounding point 1 of the aluminum electrolytic cell C1, and the voltage V1 will return to the main control computer processing unit through the voltage receiving wire. The main control computer processing unit calculates the resistance value of the insulation resistance Z1 according to the voltage V1 and compares it with the threshold value (each aluminum electrolytic cell has a threshold value requirement). If the resistance value of the insulation resistance Z1 is less than the threshold value, it means that the aluminum electrolytic cell has a fault and an alarm should be issued immediately. At the same time, the number of the aluminum electrolytic cell is displayed, thereby realizing the positioning of the aluminum electrolytic cell with a ground leakage fault.

[0069] The main control computer processing unit includes a signal generating module, a sending module, a receiving module, a calculating module, a judging module, a predictive alarm module and a controller, and the signal generating module, the sending module, the receiving module, the calculating module, the judging module and the predictive alarm module are connected to the controller respectively.

[0070] The signal generating module is used to generate the measuring signal; the sending module is used to send the measuring signal to the grounding point of the aluminum reduction cell to be measured; the receiving module is used to receive the voltage of the grounding point; the calculating module is used to calculate the insulation resistance value of the aluminum reduction cell to be measured according to the voltage of the grounding point; the judging module is used to compare the resistance value calculated by the calculating module; the predictive alarm module is used to issue an alarm prompt according to the result of the comparing module. If the insulation resistance value is less than the standard value, it means that the aluminum reduction cell has a fault and an alarm should be issued immediately. At the same time, the number of the aluminum reduction cell is displayed, so as to locate the aluminum reduction cell with a ground leakage fault.

[0071] The main control computer processing unit also includes an early warning module for predicting changes in insulation resistance based on an LSTM neural network and historical insulation resistance values ​​to the ground, thereby realizing early warning of leakage faults in the aluminum electrolysis cell and at various points in the aluminum electrolysis cell.

[0072] Based on the above system, Figure 2 As shown, the present invention also provides a method for locating and detecting earth leakage faults in an aluminum reduction cell series, including an aluminum reduction cell series earth leakage fault locating and early warning detection system, specifically comprising the following steps:

[0073] S1: When testing the aluminum electrolytic cell Ci-1 to be tested, the main control computer processing unit generates a measurement signal and sends it to the grounding point i-1 through the i-1th signal sending wire, and generates a voltage Vi-1 at the grounding point i-1;

[0074] S2: The measurement signal returns to the main control computer processing unit after passing through the i-1th insulation resistor and the aluminum electrolytic cell Ci-1 to be tested; the i-1th voltage receiving wire feeds back the voltage Vi-1 to the main control computer processing unit, and the main control computer processing unit simultaneously receives the voltage values ​​of the grounding points of other aluminum electrolytic cells.

[0075] In this embodiment, the grounding point of each aluminum electrolysis cell is connected through a grounding grid. Therefore, when the voltage Vi-1 is generated at the grounding point i-1 of the aluminum electrolysis cell to be tested, the grounding points of other aluminum electrolysis cells will also generate corresponding voltages and transmit them to the main control computer processing unit for storage through corresponding voltage receiving wires.

[0076] S3: The main control computer processing unit calculates the resistance value of the i-1th insulation resistor according to the received voltage and compares it with the threshold value. If the insulation resistance value is less than the threshold value, it means that the aluminum reduction cell has a ground leakage fault and an alarm should be given immediately. At the same time, the number of the aluminum reduction cell is displayed, thereby locating the aluminum reduction cell with the ground leakage fault.

[0077] In this embodiment, the resistance value of the i-1th insulation resistor is calculated as follows: Figure 3 As shown, this is the equivalent circuit diagram of the insulation resistance:

[0078] The main control computer processing unit sends out a frequency point FSK signal, and generates a voltage Vi-1 at the grounding point i-1 through the current limiting resistor ROi-1 and the signal sending wire. At the same time, other grounding points will also generate corresponding voltages through the grounding grid. For the convenience of calculation, only the influence of the two adjacent front and rear aluminum reduction cells on the aluminum reduction cell to be tested is considered. The other aluminum reduction cells are far apart, and the influence on the aluminum reduction cell to be tested can be ignored.

[0079] S3-1: Calculate the resistance RG between the grounding point i-1 of the i-1th insulation resistance and the adjacent grounding point i-2 a And the resistance RG between ground point i-1 and the adjacent ground point i b :

[0080]

[0081] In formula (1), RG a Represents the resistance between ground point i-1 and ground point i-2; RG b Represents the resistance between ground point i-1 and ground point i; V i-1 Represents the voltage of ground point i-1; V i-2 Represents the voltage of ground point i-2; V in Indicates the measured signal voltage; V i represents the voltage at grounding point i; R0 represents the resistance value of the current limiting resistor ROi-1.

[0082] S3-2: Calculate the resistance value Z of the insulation resistance Zi-2 and Zi adjacent to the i-1 insulation resistance i-2 , Z i :

[0083]

[0084] In formula (2), Z i-2 represents the resistance value of the i-2nd insulation resistor Zi-2 adjacent to the i-1th insulation resistor, Z i represents the resistance value of the i-th insulation resistor Zi adjacent to the i-1-th insulation resistor;

[0085] S3-3: Calculate the resistance value Z of the i-1th insulation resistor Zi-1 i-1 :

[0086] Calculate first Ra=RG a +Z i-2 , Rb=RG b +Z i ,

[0087] Then let R ab =R a / / R b , because R z =R ab / / Z i-1

[0088] Then we can get:

[0089] Solve to get the resistance value Z of the i-1th insulation resistor Zi-1 i-1 :

[0090] In formula (3), / / represents the resistance parallel operator, R z Represents the total equivalent resistance of the i-1th grounding point. Ra and Rb are intermediate variables introduced for the convenience of calculation.

[0091] In this embodiment, when measuring the first aluminum electrolytic cell, i.e. calculating the first insulation resistance, the previous value is considered to be infinite, and Ra in formula (3) is infinite, and R ab =R b ; When measuring the last aluminum electrolytic cell, i.e. calculating the i-th insulation resistance, its next value is considered infinite, corresponding to R in formula (3) b is infinite, R ab =Ra.

[0092] In this embodiment, for those aluminum reduction cells that have not yet experienced ground leakage faults, the main control computer processing unit can use a neural network method to predict the trend of insulation resistance value changes based on historical insulation resistance values. The neural network method is implemented using the "Long Short Term Memory Neural Network" LSTM. The LSTM neural network is suitable for processing and predicting important events with very long intervals and delays in time series. Because the change in the insulation resistance of the aluminum reduction cell to the ground is a long-term and gradually deteriorating process, which is consistent with the object applicable to the LSTM method, the LSTM method can be used to well predict the trend of insulation resistance changes of the aluminum reduction cell, and the early warning success rate is high.

[0093] It also includes S4: using LSTM neural network to calculate the historical insulation resistance value of each aluminum electrolytic cell collected by the main control computer processing unit, and output the predicted value of the insulation resistance value. Figure 4 As shown,

[0094] S4-1: Construct an LSTM-based aluminum electrolysis cell insulation resistance prediction network;

[0095] In this embodiment, since the LSTM network has long-term memory for data, the time correlation of the data is fully considered, and the contribution of historical information to the current prediction can be automatically adjusted according to the current state. Therefore, a prediction network for the insulation resistance of aluminum electrolytic cell to ground based on LSTM is constructed, such as Figure 5 shown.

[0096] S4-2: Obtain the historical insulation resistance values ​​of the aluminum electrolytic cell stored in the main control computer processing unit and divide them into a training set, a validation set and a test set;

[0097] In this embodiment, the historical insulation resistance values ​​of the aluminum reduction cell are obtained from the storage module of the main control computer processing unit, the historical insulation resistance values ​​are normalized, and divided into a training set, a verification set, and a test set in a ratio of 8:1:1.

[0098] In this embodiment, the historical insulation resistance value of any aluminum electrolytic cell is collected once every time period t, and collected n times in total as the model input, and the corresponding resistance value at time T is used as the label. The relationship between the predicted value and the set threshold is used to determine whether the aluminum electrolytic cell has ground leakage.

[0099] S4-3: Initialize the parameters of the LSTM-based aluminum electrolysis cell insulation resistance prediction network model:

[0100] The network output y and the initial value c <t-1>< / t-1> Initialize to 0; the weight matrix w of the network hidden layer c and bias b c , the weight matrix w of the input gate i and bias b i, update the gate weight matrix w u and bias b u , the weight matrix w of the output gate o and bias b o And the weight matrix w of the forget gate f and bias b f Initialize to a random number between 0 and 1; set the number of neurons in the network input layer to i+1, and the number of neurons in the network hidden layer, input gate, forget gate, and output gate to i; c <t>< / t> Represents the memory information at time t; a <t>< / t> represents the output of the model at time t; x <t>< / t> Represents the input of the model at time t.

[0101] S4-4: Training of the LSTM-based aluminum electrolysis cell insulation resistance prediction network model:

[0102] Set the hidden layer neuron activation function A, output unit activation function B, model loss function L, reverse tuning algorithm J and target error value E; use the divided sample training set to train the model, and use the reverse tuning algorithm J to optimize the model weights and biases according to the difference between the predicted value and the true value.

[0103] In a single LSTM unit structure, the forget gate Γ is used f <t>< / t> According to the output a at time t-1 <t-1>< / t-1> and the input x at time t <t>< / t> Calculate the amount of information inherited from the previous time step:

[0104] Γ f <t>< / t> =σ(W f [a <t-1>< / t-1> ,x <t>< / t> ]+b f );

[0105] At the same time according to a <t-1>< / t-1> and x <t>< / t> Calculate the memory information that needs to be updated Update gate Γ u <t>< / t> calculate The proportion of node memory information, the final node memory information c <t>< / t> Determined by the forget gate and the update gate:

[0106]

[0107] Γ u <t>< / t> =σ(W u [a <t-1>< / t-1> ,x <t>< / t> ]+b u ),

[0108]

[0109] Then the model output is determined by the output gate Γ o <t>< / t> Calculated from node memory information

[0110] Γ o <t>< / t> =σ(W o [a <t-1>< / t-1> ,x <t>< / t> ]+b o )

[0111] a <t>< / t> =Γ o <t>< / t> ·tanh(c <t>< / t> );σ represents the sigmoid function;

[0112] S4-5: Optimize the LSTM-based aluminum electrolysis cell insulation resistance prediction network model:

[0113] The validation set is brought into the trained model for performance verification, and the model hyperparameters are fine-tuned according to the model bias and variance; the model performance is finally evaluated using the test set, and the mean absolute error evaluation method is used as the LSTM performance evaluation indicator.

[0114] S4-6: Input the collected insulation resistance values ​​of the aluminum electrolytic cell into the trained model to obtain insulation resistance prediction results.

[0115] Those skilled in the art will appreciate that the above-mentioned embodiments are specific examples for implementing the present invention, and in actual applications, various changes may be made thereto in form and detail without departing from the spirit and scope of the present invention.

Claims

1. A method for locating and early warning the ground leakage fault of an aluminum electrolytic cell series, characterized in that: Applied to the ground leakage fault location and early warning detection system of aluminum electrolytic cell series; The detection system includes a voltage source, a DC bus, and a series of aluminum electrolytic cells connected in series through the DC bus, each of which is connected to a ground leakage fault detection circuit, and the ground leakage fault detection circuit includes a signal sending wire, a voltage receiving wire, an insulation resistor and a main control computer processing unit; One end of the signal transmission wire is connected to the main control computer processing unit, and the other end of each signal transmission wire is connected to the grounding point of the corresponding aluminum electrolytic cell; the grounding point is also connected to the corresponding aluminum electrolytic cell through an insulation resistor; the grounding point is also connected to one end of the voltage receiving wire, and the other end of the voltage receiving wire is connected to the main control computer processing unit, and the main control computer processing unit calculates the resistance value of the insulation resistor, thereby locating the aluminum electrolytic cell with a ground leakage fault; The specific steps include: S1: When testing the aluminum electrolytic cell Ci-1 to be tested, the main control computer processing unit generates a measurement signal and sends it to the grounding point i-1 through the i-1th signal sending wire, and generates a voltage Vi-1 at the grounding point i-1; S2: The measurement signal returns to the main control computer processing unit after passing through the i-1th insulation resistor and the aluminum electrolytic cell Ci-1 to be tested; the i-1th voltage receiving wire feeds back the voltage Vi-1 to the main control computer processing unit, and the main control computer processing unit simultaneously receives the voltage of the grounding point of other aluminum electrolytic cells; S3: The main control computer processing unit calculates the resistance value of the i-1th insulation resistor according to the received voltage and compares it with the threshold value. If the insulation resistance value is less than the threshold value, it means that the aluminum reduction cell has a fault and an alarm should be given immediately. At the same time, the number of the aluminum reduction cell is displayed, so as to locate the aluminum reduction cell with the ground leakage fault; The resistance value of the i-1th insulation resistor is calculated as: S3-1: Calculate the resistance RG between the grounding point i-1 of the i-1th insulation resistance and the adjacent grounding point i-2 a And the resistance RG between ground point i-1 and the adjacent ground point i b : In formula (1), RG a Represents the resistance between ground point i-1 and ground point i-2; RG b Represents the resistance between ground point i-1 and ground point i; V i-1 Represents the voltage of ground point i-1; V i-2 Represents the voltage of ground point i-2; V in Indicates the measured signal voltage; V i represents the voltage at the grounding point i; R0 represents the resistance value of the current limiting resistor; S3-2: Calculate the resistance value Z of the insulation resistance Zi-2 and Zi adjacent to the i-1 insulation resistance i-2 , Z i : In formula (2), Z i-2 represents the resistance value of the i-2nd insulation resistor Zi-2 adjacent to the i-1th insulation resistor, Z i represents the resistance value of the i-th insulation resistor Zi adjacent to the i-1-th insulation resistor; S3-3: Calculate the resistance value Z of the i-1th insulation resistor Zi-1 i-1 : Calculate first Ra=RG a +Z i-2 , Rb=RG b +Z i , Then let R ab =R a / / R b , because R z =R ab / / Z i-1 , Then we can get: Solve to get the resistance value Z of the i-1th insulation resistor Zi-1 i-1 : In formula (3), / / represents the resistance parallel operator, R z The total equivalent resistance Ra and Rb representing the i-1th grounding point are intermediate variables introduced for the convenience of calculation.

2. The method for locating and early warning the ground leakage fault of an aluminum reduction cell series according to claim 1, characterized in that: Including S4: using LSTM neural network to calculate the historical insulation resistance value of each aluminum electrolytic cell collected by the main control computer processing unit, and output the predicted value of the insulation resistance.

3. The method for locating and early warning the ground leakage fault of the aluminum electrolytic cell series according to claim 2, characterized in that: The calculation of the predicted value of the insulation resistance comprises the following steps: S4-1: Construct an LSTM-based aluminum electrolysis cell insulation resistance prediction network; S4-2: Obtain the historical insulation resistance values ​​of the aluminum electrolytic cell stored in the main control computer processing unit and divide them into a training set, a validation set and a test set; S4-3: Initialize the parameters of the LSTM-based aluminum electrolysis cell insulation resistance prediction network model: The network output y and the initial value c <t-1〉 Initialize to 0; the weight matrix w of the network hidden layer c and bias b c , the weight matrix w of the input gate i and bias b i , update the gate weight matrix w u and bias b u , the weight matrix w of the output gate o and bias b o And the weight matrix w of the forget gate f and bias b f Initialize to a random number between 0 and 1; set the number of neurons in the network input layer to i+1, and the number of neurons in the network hidden layer, input gate, forget gate, and output gate to i; c t〉 represents the memory information at time t; a t> represents the output of the model at time t; x t> Represents the input of the model at time t; S4-4: Training of the LSTM-based aluminum electrolysis cell insulation resistance prediction network model: The model is trained using the divided sample training set, and the model weights and biases are optimized using the reverse tuning algorithm J according to the difference between the predicted value and the true value; In a single LSTM unit structure, the forget gate Γ is used f t> According to the output a at time t-1 t-1> and the input x at time t t> Calculate the amount of information inherited from the previous time step: C f t> =s ( W f [a t-1 ,x t ]+b f ) ; At the same time according to a <t-1>< / t-1> and x <t>< / t> Calculate the memory information that needs to be updated Update gate Γ u <t〉 calculate The proportion of node memory information, the final node memory information c 〈t〉 Determined by the forget gate and the update gate: C u t> =s ( W u [a t-1 ,x t ]+b u ) , Then the model output is determined by the output gate Γ o t〉 Calculated from node memory information C o t〉 =s ( W o [a t-1 ,x t ]+b o ) ,a t〉 =C o t〉 ·tanh(c t> ),σ represents a sigmoid function; S4-5: Input the collected insulation resistance value of the aluminum electrolytic cell into the trained model to obtain a predicted value of the insulation resistance.

4. The method for locating and early warning the ground leakage fault of an aluminum electrolytic cell series according to claim 1, characterized in that: Each aluminum electrolysis cell has a unique number, and each number corresponds to a unique installation location.

5. The method for locating and early warning the ground leakage fault of an aluminum electrolytic cell series according to claim 1, characterized in that: The signal sending wire is connected to the main control computer processing unit through a current limiting resistor.

6. The method for locating and early warning the ground leakage fault of an aluminum electrolytic cell series according to claim 1, characterized in that: The main control computer processing unit includes a signal generating module, a sending module, a receiving module, a calculating module, a judging module, a predicting alarm module and a controller, and the signal generating module, the sending module, the receiving module, the calculating module, the judging module and the predicting alarm module are connected to the controller respectively; A signal generating module is used to generate a measuring signal; a sending module is used to send the measuring signal to the grounding point of the aluminum electrolytic cell to be measured; A receiving module, used for receiving the voltage of the grounding point; The calculation module is used to calculate the resistance value of the insulation resistance of the aluminum electrolytic cell to be tested according to the voltage of the grounding point; the judgment module is used to compare the resistance value calculated by the calculation module; the prediction alarm module is used to issue an alarm prompt according to the result of the comparison module.

7. The method for locating and early warning the ground leakage fault of an aluminum reduction cell series as claimed in claim 6, characterized in that: The measurement signal generated by the signal generating module is a sinusoidal multi-carrier FSK signal, and the frequency range of the measurement signal is 500 Hz to 5 KHz.

8. The method for locating and early warning the ground leakage fault of an aluminum electrolytic cell series as claimed in claim 1, characterized in that: The main control computer processing unit also includes an early warning module, which is used to predict the change of insulation resistance based on the LSTM neural network and historical insulation resistance values, so as to realize the early warning of the ground leakage fault of the aluminum electrolytic cell.

Images