JP2023076312A - Mechanical learning device and storage battery state determination device - Google Patents

Abstract

To provide a mechanical learning device and a storage battery state determination device that allow an inspection person to make an accurate determination regardless of the experience or the ability of the inspection person.SOLUTION: A mechanical learning device 10 includes: a first input data acquisition unit 111; a first label acquisition unit 112; and a first learning model construction unit 113. The first input data acquisition unit 111 acquires actual object picture information of at least one of the appearance and the inside of a storage battery 40 as input data. The first label acquisition unit 112 acquires actual object data showing the state of the storage battery 40 corresponding to the actual object picture information, as a label. The first learning model construction unit 113 performs a learning with a teacher by using a combination of the actual object picture information and actual object data as teacher data, thereby constructing a first learning model as a learning model to determine the state of the storage battery 40.SELECTED DRAWING: Figure 2

Description

The present invention relates to a machine learning device and a storage battery state determination device.

2. Description of the Related Art Conventionally, techniques for monitoring the state of batteries in electronic devices are known. Patent document 1 describes this type of technology. In Patent Document 1, a battery management system includes an electronic device having a battery and a management server connected to the electronic device via a network, and the electronic device stores battery information including a state indicating the state of the battery. A technology in which information is stored in a storage medium, the management server predicts the current state of the battery of the electronic device based on the battery information of the electronic device, and based on the predicted result, control information for battery control of the electronic device is sent to the electronic device. is described.

Japanese Patent Application No. 2020-523219

By the way, in the maintenance and inspection of a storage battery during floating charging, there are cases where the external appearance of the container and lid and the state of the positive and negative electrode plates inside the storage battery are visually determined. Visual judgment requires experience and ability, and there is room for improvement in the accuracy of judgment when the inspector lacks experience and ability.

An object of the present invention is to provide a machine learning device and a storage battery state determination device that enable more accurate determination regardless of the experience and ability of an inspector.

(1) A machine learning device according to the present invention includes an input data acquisition unit that acquires image data of at least one of the exterior and interior of a storage battery as input data, and state information indicating the state of the storage battery corresponding to the image data. as a label, and a model construction unit that constructs a learning model for determining the state of the storage battery by performing supervised learning using a set of the image data and the state information as teacher data. , provided.

(2) In the machine learning device according to the present invention, the label acquisition unit acquires degree information indicating the degree of abnormality as the state information, and the model construction unit obtains the degree information according to input image data. Build a learning model to output.

(3) In the machine learning device according to the present invention, the label acquisition unit acquires, as the state information, abnormality position information indicating an abnormality occurrence position in the storage battery corresponding to the image data, and the model construction unit obtains the By performing supervised learning using pairs of image data and the abnormal position information as teacher data, a learning model for determining the abnormal position is constructed.

(4) In the machine learning device according to the present invention, the input data acquisition unit acquires the specific gravity information of the electrolyte at a predetermined timing before the end of the life of the storage battery as input data, and the label acquisition unit acquires the predetermined remaining life period information indicating the remaining life of the storage battery from the timing of until the end of the life of the storage battery is acquired as the state information, and the model construction unit creates a set of the specific gravity information and the remaining life period information A learning model for determining the remaining life is constructed by performing supervised learning as teacher data.

(5) A storage battery state determination device according to the present invention includes a learning model acquisition unit that acquires the learning model constructed by the machine learning device according to any one of (1) to (4), and an external and internal view of the storage battery. a new input data acquisition unit that acquires at least one of image data or specific gravity information of the electrolyte of the storage battery as new input data; and determination of the state of the storage battery based on the new input data and the learning model. and a storage battery state determination unit configured to output a determination result.

According to the present invention, it is possible to provide a machine learning device and a storage battery state determination device that enable more accurate determination regardless of the experience and ability of an inspector.

BRIEF DESCRIPTION OF THE DRAWINGS It is a figure which shows the whole storage battery management system structure which concerns on embodiment of this invention. FIG. 3 is a functional block diagram of a first learning section provided in the machine learning device according to the embodiment of the present invention; FIG. 4 is a functional block diagram of a second learning section provided in the machine learning device according to the embodiment of the present invention; FIG. 4 is a functional block diagram of a third learning section provided in the machine learning device according to the embodiment of the present invention; 1 is a functional block diagram of a storage battery state determination device according to an embodiment of the present invention; FIG. 4 is a flow chart showing the operation of the first learning unit among the operations of the machine learning device according to the embodiment of the present invention; 8 is a flow chart showing the operation of the second learning unit among the operations of the machine learning device according to the embodiment of the present invention; 8 is a flow chart showing the operation of the third learning unit among the operations of the machine learning device according to the embodiment of the present invention; It is a flow chart which shows operation of a storage battery state judging device concerning an embodiment of the present invention.

An embodiment of the present invention will be described below with reference to FIGS. 1 to 5. FIG.

[1 Configuration of Embodiment]
[1.1 Overall configuration]
First, the configuration of the storage battery management system S according to this embodiment will be described. FIG. 1 is a diagram showing the overall configuration of a storage battery management system. The storage battery management system S is a system for managing the state of a plurality of floating charging storage batteries 40 provided in, for example, a power plant.

The storage battery 40 according to this embodiment is composed of a plurality of batteries called cells, and has six cells 40a, 40b, 40c, 40d, 40e, and 40f as shown in FIG. Note that the number of cells is not limited to this and can be freely set. The storage battery management system S includes a machine learning device 10, a storage battery state determination device 20, a terminal device 30, and a network N, as shown in FIG. Note that the number of the machine learning device 10, the storage battery state determination device 20, and the terminal device 30 may be one or more.

Further, the machine learning device 10, the storage battery state determination device 20, and the terminal device 30 are each connected to the network N, and can communicate with each other via the network N. The network N is, for example, a LAN (Local Area Network), the Internet, a public telephone network, or a combination thereof. The specific communication method in the network N, whether it is a wired connection or a wireless connection, etc., is not particularly limited. Note that the machine learning device 10 and the storage battery state determination device 20 may be directly connected via a connection unit instead of communication using the network N.

The machine learning device 10 includes a first learning unit 11, a second learning unit 12, and a third learning unit 13, as shown in FIG. The first learning unit 11, the second learning unit 12, and the third learning unit 13 are learning models that perform supervised learning based on input teacher data and are used by the storage battery state determination device 20 to determine the state of the storage battery. .

In the present embodiment, actual photograph information obtained by photographing the exterior or interior of the actual storage battery 40 as input data is used as teacher data for constructing a learning model capable of determining abnormality of the storage battery 40 from actual photographs. . In addition, abnormality determination result information as status information of the storage battery 40 determined by a skilled inspector as a label is used as the teacher data. That is, as the training data, a set of data of actual photograph information and abnormality determination result information is used.

Further, in the present embodiment, the above-described actual photograph information is used as input data for training data for constructing a learning model capable of determining the abnormal position of the storage battery 40 from the actual photograph. In addition, abnormal position information as status information of the storage battery 40 determined by a skilled inspector as a label is used for the teacher data. That is, data of a set of actual photograph information and abnormal position information is used as the training data.

Further, in the present embodiment, the teacher data for constructing a learning model capable of determining the remaining life of the storage battery 40 from the actual data of the storage battery 40 includes the actual data such as the voltage of the storage battery 40 whose life has ended as input data. history information is used. The actual data according to the present embodiment includes measurement data such as the liquid level and temperature of the electrolyte in the storage battery 40, specific gravity information, voltage, outside temperature, etc. measured at the time of patrol inspection, aging information from the start date of use, manufacturing date and other equipment data. The history information includes actual data from the start of use of the storage battery 40 to the end of its life.

The information contained in the actual data is not limited to this, but includes, for example, the rate of decrease of the electrolyte, the state information of the positive and negative plates, the variation information of the electrolyte temperature and voltage between cells, the storage battery capacity, and the storage battery deterioration. It may be the charge/discharge efficiency of a storage battery, or the like.

In addition, state information of the storage battery 40 whose life has ended as a label and life information as a remaining life span are used as the teacher data. That is, as the teacher data, a set of data of the actual data and the life information of the storage battery 40 is used. The history information included in the teacher data includes inspection date and time information, and the remaining life at the time of measurement of the measurement data can be calculated based on the life information and used for machine learning.

The storage battery state determination device 20 acquires new input data necessary for determining the state of the storage battery 40 and applies it to the learning model constructed by the machine learning device 10, thereby determining the state of various storage batteries 40. is. Further, in the present embodiment, the storage battery state determination device 20 acquires new input data for each cell and performs state determination of the storage battery 40 .

The terminal device 30 is a device that displays battery state determination result information output from the battery state determination device 20 and outputs input from the user of the terminal device 30 to the battery state determination device 20 . The terminal device 30 is realized by, for example, a mobile phone such as a smart phone or a mobile terminal such as a tablet terminal, but is not limited to this. Also, in FIG. 1, the terminal device 30 is shown as a plurality of terminal devices 30a to 30q, but is not limited to this, and for example, only one device may exist.

In addition, in the storage battery management system S according to the present embodiment, the machine learning device 10 and the storage battery state determination device 20 are provided separately, but they may be provided in the same device.

[1.2 Configuration of machine learning device]
As described above, the machine learning device 10 includes the first learning section 11 , the second learning section 12 and the third learning section 13 . The first learning unit 11 according to the present embodiment constructs a first learning model for determining abnormality of the storage battery 40 included in new actual photograph information including the storage battery 40 as a subject. Abnormality of the storage battery 40 indicates a state in which the container, lid, positive/negative plates, or electrolytic solution of the storage battery 40 is cracked, deformed, damaged, or the like.

FIG. 2 is a functional block diagram of the first learning unit 11. As shown in FIG. As shown in FIG. 2, the first learning unit 11 includes a first input data acquisition unit 111, a first label acquisition unit 112, a first learning model construction unit 113 as a model construction unit, and a first learning model storage unit 114. Prepare.

The first input data acquisition unit 111 acquires actual photograph information including the storage battery 40 as an object as input data. For example, the first input data acquisition unit 111 executes a process of acquiring information on a storage battery to be inspected that is input to the terminal device 30 and transmitted to the machine learning device 10 via the network N.

The first label acquisition unit 112 acquires, as a label, the level of abnormality (degree of abnormality) of the storage battery 40 included in the actual photograph information including the storage battery 40 as a subject. The setting of the abnormality level of the storage battery 40 is based on, for example, abnormality determination result information as information obtained by a skilled inspector who directly looks at a photograph of the actual product or the actual product. However, it is not limited to this.

In the present embodiment, the determination of abnormality of the storage battery 40 is made in four stages of degrees (levels) from level A to level D shown below. The level of abnormality of the storage battery 40 may be, for example, a score indicated by a numerical value, and the degree of abnormality is determined stepwise using alphabets or symbols such as excellent, good, acceptable, and unacceptable. may

Level A is an abnormal state in which an abnormality has occurred in the container or lid of the storage battery 40, the positive/negative plates inside the storage battery 40, or the electrolytic solution, and immediate countermeasures are required. For example, at level A, the battery case of the storage battery 40 has a crack with an opening that allows the electrolyte to leak out, and a large deformation that is considered to be caused by abnormal heat generation can be confirmed. This is a level that can be confirmed from the actual photograph. Alternatively, level A is a situation in which normal operation cannot be performed, such as the positive and negative plates inside the storage battery 40 being damaged. Alternatively, level A is a state in which normal operation cannot be performed, such as the electrolyte being remarkably degraded or the liquid level being abnormally reduced.

Level B is a state in which an inspection of the storage battery 40 is required at the next inspection. For example, level B is a state in which, although leakage of electrolyte or the like cannot be confirmed in the storage battery 40, cracks or deformation can be confirmed on the surface, and there is a high possibility that the storage battery 40 will eventually break. Alternatively, level B is a state in which there is a high degree of damage to the internal positive and negative plates, and there is a high possibility that the device will eventually become inoperable. Alternatively, level B is a state in which there is a high possibility that operation will eventually become impossible, such as confirmation of deterioration in the electrolyte, reduction in the liquid level, or the like.

Level C is a state in which it is necessary to reduce the frequency of inspection of the storage battery 40 . For example, it is necessary to increase the frequency of inspection of the determined parts from once a year to three times a year. At level C, for example, the surface of the storage battery 40 cannot be confirmed to have electrolyte leakage, cracks, deformation, or damage, but it can be confirmed that the external shape of the storage battery 40 is deformed from a normal storage battery, and deterioration of the storage battery 40, etc. It is a state in which there is a possibility that Alternatively, level C is a state in which damage, deformation, deterioration, or the like can be confirmed in the positive and negative plates inside, and there is a possibility that deterioration or the like has occurred. Alternatively, level C is a state in which there is a possibility that deterioration or the like has occurred, such as a slight change in the electrolytic solution such as deterioration or a decrease in the liquid level being confirmed.

Level D is a state in which the surface and inside of the container and lid of the storage battery 40 are the same as those of a normal storage battery, and the storage battery 40 is in a state in which no abnormality has occurred.

The first learning model construction unit 113 performs supervised learning using a set of the first input data and the first label as teacher data, thereby performing first learning for determining abnormality of the storage battery 40 included in the actual photograph information. build a model; The constructed first learning model is used for determination by the storage battery state determination device 20 .

The first learning model construction unit 113 according to this embodiment constructs a first learning model as a learned model by machine learning. For example, the first learning model construction unit 113 constructs the first learning model by performing deep learning using a multi-layer (input layer, output layer, intermediate layer) neural network.

Note that the algorithm used by the first learning model construction unit 113 to construct the first learning model is not limited to this. For example, the first learning model building unit 113 may perform supervised learning using a support vector machine (also referred to as SVM) to build the first learning model, or may use other algorithms. may be used for learning to build a first learning model.

In this case, the first learning model construction unit 113 uses, as the first label, a binarized label indicating whether or not it corresponds to a specific degree of abnormality, and the space containing the input data. , a separating hyperplane is calculated so that the margin is maximized with respect to whether or not the specific degree of anomaly is met. Furthermore, the first learning model constructing unit 113 can use the coefficients of this hyperplane as parameters of a learning model that is used by the battery state determination device 20 (to be described later) to determine the state of the battery.

The first learning model storage unit 114 executes processing for storing the first learning model constructed by the first learning model construction unit 113 in the storage device of the machine learning device 10 . For example, when the first learning model construction unit 113 constructs the first learning model, the first learning model storage unit 114 stores the constructed first learning model in the storage device of the machine learning device 10. .

The second learning unit 12 constructs a second learning model for determining an abnormal position such as a crack in the storage battery 40 included in new actual photograph information including the storage battery 40 as a subject. FIG. 3 is a functional block diagram of the second learning unit 12. As shown in FIG. The second learning unit 12 includes a second input data acquisition unit 121 , a second label acquisition unit 122 , a second learning model building unit 123 as a model building unit, and a second learning model storage unit 124 .

The second input data acquisition unit 121 acquires the actual photograph including the storage battery 40 as a subject as second input data. The second label acquisition unit 122 acquires, as a second label, the determination result of an abnormal position such as a crack in the storage battery 40 included in the actual photograph information including the storage battery 40 as a subject. Further, in this embodiment, the determination result of the abnormal position is obtained by a skilled inspector directly confirming the photograph of the actual product or the actual product.

It should be noted that the determination result of the abnormal position is not limited to the visual determination by the skilled inspector. In addition, the results of this determination include not only locations where cracks have occurred, but also locations where the electrolyte is leaking from deformation locations, breakage locations, and cracks.

The second learning model construction unit 123 performs supervised learning using the set of the second input data and the second label as teacher data in the same manner as the first learning model construction unit 113, thereby including the storage battery 40 as a subject. A second learning model is constructed for determining the location of an abnormality such as a crack in the storage battery 40 included in the new physical photograph information. The constructed second learning model is used for determination by the storage battery state determination device 20 .

The second learning model storage unit 124 executes processing for storing the second learning model constructed by the second learning model construction unit 123 in the storage device of the machine learning device 10 . For example, when the second learning model building unit 123 builds the second learning model, the second learning model storage unit 124 stores the built second learning model in the storage device of the machine learning device 10. .

The third learning unit 13 constructs a third learning model for predicting the remaining life of the storage battery 40 from actual product data. As the teacher data, history information and life information of the actual data of the storage battery 40 whose life has ended as described above are used. FIG. 4 is a functional block diagram of the third learning unit 13. As shown in FIG. The third learning unit 13 includes a third input data acquisition unit 131 , a third label acquisition unit 132 , a third learning model construction unit 133 as a model construction unit, and a third learning model storage unit 134 .

The third input data acquisition unit 131 acquires the aforementioned actual data from history data in the measurement result database 213 of the storage unit 21 of the storage battery state determination device 20 described below as third input data. Further, the third input data acquisition unit 131 may calculate other data based on the actual data and acquire it as the third input data. For example, the third input data acquisition unit 131 may calculate the rate of decrease of the electrolyte solution from the water level of the electrolyte solution level acquired at another timing.

The third label acquisition unit 132 acquires the remaining life of the storage battery 40 indicated by the third input data as the third label. The remaining life of the storage battery 40 acquired by the third label acquisition unit 132 is the measurement date of the external temperature, etc. of the third input data read from the history data of the measurement result database 213, which will be described later, and the date and time when the life of the storage battery 40 ends. is calculated from

The third learning model construction unit 133 performs supervised learning using a combination of the third input data and the third label as teacher data in the same manner as the first learning model construction unit 113, thereby calculating the remaining life of the storage battery 40. Build a third learning model to predict. The constructed third learning model is used for determination by the storage battery state determination device 20 .

The third learning model storage unit 134 executes processing for storing the learning model constructed by the third learning model construction unit 133 in the storage device of the machine learning device 10 . For example, when the third learning model construction unit 133 constructs the third learning model, the third learning model storage unit 134 stores the constructed third learning model in the storage device of the machine learning device 10. .

[1.3 Configuration of storage battery state determination device]
FIG. 5 is a functional block diagram of the storage battery state determination device 20. As shown in FIG. The storage battery state determination device 20 includes a storage unit 21 , a control unit 22 , a communication unit 23 and a display unit 24 .

The storage unit 21 stores learning models acquired from the machine learning device 10 . Further, the storage unit 21 stores an equipment information database 211, a learning model database 212, and a measurement result database 213. FIG.

Equipment data and the like are stored in the equipment information database 211 . For example, the facility information database 211 stores facility information such as the date of use of the storage battery 40, the serial number, the manufacturer, and the date of manufacture.

The learning model database 212 stores a plurality of learning models transmitted from the first learning model construction unit 113, the second learning model construction unit 123, and the third learning model construction unit 133 of the machine learning device 10, and stores the state of the storage battery 40. It is read when judgment processing is executed.

The measurement result database 213 receives inspection results transmitted via the inspector's terminal device 30 during patrol inspection of the storage battery 40 . For example, the measurement result database 213 stores storage battery information such as image data of the storage battery 40, liquid level of the electrolyte, specific gravity, outside air temperature, voltage, and temperature of the electrolyte in the cell. The inspection results may be automatically measured by providing various sensors and a communication device in the storage battery 40 and transmitted to the measurement result database 213 via the network N for storage.

The control unit 22 is a part that controls the entire storage battery state determination device 20, and by reading and executing various programs as appropriate from storage areas such as ROM, RAM, flash memory, or hard disk (HDD), the present embodiment It realizes various functions in The control unit 22 may be a CPU. The control unit 22 includes an actual photo acquisition unit 221, an actual data acquisition unit 222, a learning model acquisition unit 223, an abnormality determination unit 224, an abnormal position determination unit 225, a remaining life determination unit 226, and a storage battery state management unit as a storage battery state determination unit. 227;

The actual photograph acquisition unit 221 acquires new actual photograph information separately from the actual photograph information included in the teacher data used in learning by the machine learning device 10 . For example, an inspector causes the storage battery state determination device 20 to read the actual photograph information through the terminal device 30 at the time of patrol inspection, and the actual photograph acquisition unit 221 acquires the actual photograph information.

The actual data acquisition unit 222 acquires new actual data separately from the actual data included in the teacher data used when the machine learning device 10 performs machine learning. For example, the inspector operates the storage battery state determination device 20 to read the actual product data via the terminal device 30, and the actual product data acquisition unit 222 acquires the actual product photograph information.

Note that the actual product data acquisition unit 222 obtains new actual product data from the storage battery state determination device 20 to which new actual product photograph information and actual product data are input by the inspector's operation of an input device such as a keyboard and a mouse provided in the storage battery state determination device 20 . Data or actual photograph information may be acquired. Further, the actual data acquiring unit 222 may acquire the actual data and the actual photograph information stored in the measurement result database 213 and the facility information database 211 of the storage unit 21 of the storage battery state determination device 20 as new input data. .

The learning model acquisition unit 223 executes processing for acquiring a learning model used for determination when determining the state of the storage battery. For example, when determining an abnormality in the storage battery 40 , the learning model acquisition unit 223 refers to the learning model database 212 to search, read, and acquire a first learning model for determining an abnormality in the storage battery 40 . When the learning model database 212 does not have the first learning model, the learning model acquisition unit 223 requests the machine learning device 10 to transmit the first learning model.

The abnormality determination unit 224 determines abnormality of the storage battery 40 based on the actual photograph information acquired by the actual photograph acquisition unit 221 and the first learning model acquired by the learning model acquisition unit 223 .

Based on the actual photograph information acquired by the actual photograph acquisition unit 221 and the second learning model acquired by the learning model acquisition unit 223, the abnormal position determination unit 225 determines the abnormal position of the storage battery included in the new actual photograph information. judge.

The remaining life determination unit 226 determines the remaining life of the storage battery 40 based on the actual product data acquired by the actual product data acquisition unit 222 and the third learning model acquired by the learning model acquisition unit 223 .

When the storage battery state management unit 227 acquires the storage battery state determination request, the storage battery state management unit 227 causes the actual photo acquisition unit 221 and the actual data acquisition unit 222 to acquire various types of information, and performs an abnormality determination unit 224, an abnormal position determination unit 225, and a remaining life determination. It causes the unit 226 to perform determination processing and outputs the result. For example, when the storage battery state management unit 227 acquires a storage battery state determination request received from the terminal device 30 via the communication unit 23, the storage battery state management unit 227 instructs the actual photograph acquisition unit 221 and the actual data acquisition unit 222 to acquire necessary information. . Next, the storage battery state management unit 227 causes the abnormality determination unit 224, the abnormal position determination unit 225, and the remaining life determination unit 226 to perform determination processing.

Next, the storage battery state management unit 227 transmits the result of the storage battery state determination performed by the storage battery state determination device 20 to the terminal device 30 via the communication unit 23 (to be described later), for example, information about the need for investigation and the details of the abnormality. The display unit of the terminal device 30 that has received the information displays whether or not the storage battery 40 needs to be inspected and the details of the abnormality.

The communication unit 23 is a communication interface used by the storage battery state determination device 20 to communicate with the machine learning device 10 and the terminal device 30 .

The display unit 24 is a monitor that displays an operation screen or the like for performing processing related to the storage battery management system S. FIG.

[2 Operation of Embodiment]
Next, the operation of the storage battery management system S according to this embodiment will be described. The storage battery state determination process in the storage battery management system S according to the present embodiment includes a learning stage in which a learning model is constructed by supervised learning using pairs of input data and labels acquired by the machine learning device 10 as teacher data. The training data used in the learning stage is set data of the actual photograph information or actual data of the storage battery 40 inspected by a skilled inspector and the judgment result information of the storage battery 40 inspected by the skilled inspector.

In addition, in the storage battery state determination process, there is a utilization stage in which storage battery state determination is performed by the storage battery state determination device 20 using the learning model from actual photo information or actual data as new input data input from the terminal device 30 or the like. There is The learning stage operation by the machine learning device 10 and the usage stage operation by the storage battery state determination device 20 will be described below.

[2.1 Learning model construction processing of the machine learning device 10]
First, the learning model construction processing of the machine learning device 10 will be described with reference to FIGS. 6 to 8. FIG. The machine learning device 10 causes the first learning unit 11, the second learning unit 12, and the third learning unit 13 to perform processing for building a learning model at the timing of acquiring a machine learning execution request.

FIG. 6 is a flowchart showing learning model building processing of the first learning unit 11 of the machine learning device 10 . First, the first input data acquisition unit 111 and the first label acquisition unit 112 obtain the actual photograph information as the first input data and the first label as the first label for the storage battery 40 inspected by the skilled inspector. Determination result information of the storage battery 40 is acquired (step S10). Next, the first learning model constructing unit 113 sets the acquired combination of the actual photograph information and the determination result information as teacher data (step S11). Next, the first learning model construction unit 113 performs machine learning using the teacher data (step S12).

Next, the first learning model construction unit 113 determines whether or not to end machine learning (step S13). When the first learning model construction unit 113 determines to repeat machine learning (step S13: NO), the process proceeds to step S10. Then, the machine learning device 10 repeats the same operation. On the other hand, when the first learning model construction unit 113 determines to end the machine learning (step S13: YES), the process proceeds to step S14. Note that the conditions for terminating machine learning can be arbitrarily determined. For example, machine learning may be terminated when machine learning is repeated a predetermined number of times. In addition, the accuracy of the built learning model is improved by repeating machine learning.

Next, the first learning model constructing unit 113 executes a process of transmitting the constructed learning model to the storage battery state determination device 20 (step S14), and terminates the process.

FIG. 7 is a flowchart showing learning model building processing of the second learning unit 12 of the machine learning device 10 . First, the second input data acquisition unit 121 and the second label acquisition unit 122 acquire actual photograph information to be judged by a skilled inspector as second input data and abnormal position information as a second label (step S20). Next, the second learning model constructing unit 123 sets the acquired set of the actual photograph information and the abnormal position information as teacher data (step S21). Next, the second learning model construction unit 123 performs machine learning using the teacher data (step S22).

Next, the second learning model construction unit 123 determines whether or not to end machine learning (step S23). When the second learning model construction unit 123 determines to repeat machine learning (step S23: NO), the process proceeds to step S20. Then, the machine learning device 10 repeats the same operation. On the other hand, when the second learning model construction unit 123 determines to end the machine learning (step S23: YES), the process proceeds to step S24. Note that the conditions for terminating machine learning can be arbitrarily determined. For example, machine learning may be terminated when machine learning is repeated a predetermined number of times.

FIG. 8 is a flowchart showing learning model building processing of the third learning unit 13 of the machine learning device 10 . First, the third input data acquisition unit 131 and the third label acquisition unit 132 acquire actual product data as third input data and remaining life determination result information as a third label (step S30). Next, the third learning model construction unit 133 sets a set of the acquired actual product data and remaining life determination result information as teacher data (step S31). Next, the third learning model construction unit 133 performs machine learning using the teacher data (step S32).

Next, the third learning model construction unit 133 determines whether or not to end machine learning (step S33). When the third learning model construction unit 133 determines to repeat machine learning (step S33: NO), the process proceeds to step S30. Then, the machine learning device 10 repeats the same operation. On the other hand, when the third learning model construction unit 133 determines to end the machine learning (step S33: YES), the process proceeds to step S34. Note that the conditions for terminating machine learning can be arbitrarily determined. For example, machine learning may be terminated when machine learning is repeated a predetermined number of times.

When the storage battery state determination device 20 receives each learning model transmitted from the machine learning device 10 through the process, the storage battery state determination device 20 performs a process of storing it in the learning model database 212 of the storage device. Each learning model stored in the learning model database 212 is utilized in the storage battery state determination in the operation of the storage battery state determination device 20 described below. The machine learning device 10 can also perform further machine learning on the learning model when new teacher data is acquired.

[2.2 Storage battery state determination processing of storage battery state determination device 20]
Next, the storage battery state determination processing of the storage battery state determination device 20 will be described with reference to FIG. 9 . The storage battery state determination device 20 executes a storage battery state determination process when the storage battery state determination request is acquired.

FIG. 9 is a flowchart showing battery state determination processing of the battery state determination device 20 according to the embodiment of the present invention. First, the control unit 22 of the storage battery state determination device 20 acquires one or both of the actual photograph information and the actual data as new input data (step S40). Next, the control unit 22 confirms whether or not the new input data includes actual photograph information including the storage battery 40 as a subject (step S41).

If the new input data does not contain the actual photograph information (step S41: NO), the control section 22 shifts the process to step S47. On the other hand, if the new input data includes the actual photograph information (step S41: YES), the control unit 22 refers to the learning model database 212 and acquires the first learning model (step S42). Next, the control unit 22 determines whether the storage battery 40 is abnormal based on the actual photograph information included in the new input data acquired in step S40 and the first learning model (step S43).

Next, the control unit 22 confirms whether or not there is an abnormality in the storage battery 40 (step S44). If there is no abnormality in the storage battery 40 (step S44: NO), the control unit 22 causes the process to proceed to step S47. On the other hand, if there is an abnormality in the storage battery 40 (step S44: YES), the control unit 22 refers to the learning model database 212 and acquires the second learning model (step S45). Next, the control unit 22 determines the abnormal position of the storage battery 40 based on the acquired actual photograph information and the second learning model (step S46).

Next, the control unit 22 refers to the learning model database 212 and acquires the third learning model (step S47). Next, the control unit 22 acquires the actual data included in the new input data acquired in step S40, and determines the remaining life of the storage battery 40 based on the actual data and the third learning model (step S48). Next, the control unit 22 transmits the result determined in this process to the terminal device 30 (step S49), and terminates the process.

The machine learning device 10 configured as described above includes a first input data acquisition unit 111 that acquires, as input data, actual photograph information of at least one of the exterior and interior of the storage battery 40, A learning model for determining the state of the storage battery 40 is obtained by performing supervised learning using a first label acquisition unit 112 that acquires data as a label and a set of actual photograph information and actual data as teacher data. and a first learning model construction unit 113 that constructs a model.

As a result, the machine learning device 10 according to the present embodiment enables more accurate determination regardless of the experience and ability of the inspector.

Also, the first label obtaining unit 112 obtains degree information indicating the degree of abnormality, and the first learning model building unit 113 builds a learning model that outputs degree information according to input image data.

As a result, the machine learning device 10 according to the present embodiment can more accurately determine the abnormality of the storage battery regardless of the experience and ability of the inspector.

In addition, in the machine learning device 10 according to the present embodiment, the second label acquisition unit 122 acquires, as state information, abnormality location information indicating the location of an abnormality in the storage battery 40 corresponding to the field photo information, and performs second learning. The model construction unit 123 constructs a second learning model as a learning model for judging an abnormality occurrence position by performing supervised learning using a set of actual photograph information and abnormality position information as teacher data.

As a result, the machine learning device 10 according to the present embodiment can easily identify the location where the abnormality occurs, and can more efficiently deal with the abnormality.

Further, in the machine learning device 10 according to the present embodiment, the third input data acquisition unit 131 acquires the specific gravity information of the electrolytic solution at a predetermined timing before the end of the life of the storage battery 40 as input data, and acquires the third label. The unit 132 acquires remaining life period information indicating the remaining life of the storage battery 40 from a predetermined timing to the end of the life of the storage battery 40 as state information. A third learning model is constructed as a learning model for determining the remaining life by performing supervised learning using pairs of information as teacher data.

As a result, with the machine learning device 10 according to the present embodiment, even an inexperienced inspector can more accurately determine the remaining life of the storage battery 40, and efficient operation of the storage battery 40, such as using it as long as it has its life, is possible. can.

[3 Modifications]
Although the embodiments of the present invention have been described above, the present invention is not limited to the above-described embodiments. Moreover, the effects described in the present embodiment are merely enumerations of the most suitable effects resulting from the present invention, and the effects of the present invention are not limited to those described in the present embodiment.

In the storage battery management system S according to the present embodiment, the determination result is output after the storage battery state is determined using the learning model. It may be weighted based on the equipment information. For example, the remaining life information of the storage battery included in the determination result information may shorten or extend the remaining life based on the material used for the storage battery stored in the equipment information database 211 .

In the storage battery management system S according to the present embodiment, measurement data and the like of the storage battery are transmitted from the terminal device 30 to the storage battery state determination device 20 by the input operation of the inspector. 20 may be obtained directly from the storage battery 40 via the network. In this case, the storage battery 40 includes a communication unit that can communicate with the storage battery state determination device 20 via a network, a measurement unit that acquires data from various measurement devices that the storage battery 40 has, and measurements acquired by the measurement unit via the communication unit. A control unit that performs processing for transmitting data to the storage battery state determination device 20 is provided.

The series of processes described above can be executed by hardware or by software. In other words, the functional configurations of FIGS. 1-4 are merely examples and are not particularly limiting. That is, it is sufficient if the machine learning device 10 has a function capable of executing the above-described series of processes as a whole, and what kind of functional blocks are used to realize this function are particularly shown in the examples of FIGS. Not limited.

Also, one functional block may be composed of hardware alone, software alone, or a combination thereof. The functional configuration in this embodiment is realized by a processor that executes arithmetic processing, and processors that can be used in this embodiment are composed of various single processing units such as single processors, multiprocessors, and multicore processors. In addition, it includes a combination of these various processing devices and a processing circuit such as ASIC (Application Specific Integrated Circuit) or FPGA (Field-Programmable Gate Array).

When a series of processes is to be executed by software, a program constituting the software is installed in a computer or the like from a network or a recording medium. The computer may be a computer built into dedicated hardware. The computer may also be a computer capable of executing various functions by installing various programs, such as a general-purpose personal computer.

A recording medium containing such a program is not only constituted by a removable medium that is distributed separately from the main body of the device in order to provide the program to the user, but is also provided to the user in a state pre-installed in the main body of the device. It consists of a recording medium, etc. Removable media are composed of, for example, magnetic disks (including floppy disks), optical disks, or magneto-optical disks. Optical discs are composed of, for example, CD-ROMs (Compact Disk-Read Only Memory), DVDs (Digital Versatile Disks), Blu-ray (registered trademark) Discs, and the like. The magneto-optical disk is composed of an MD (Mini-Disk) or the like. Further, the recording medium provided to the user in a state of being pre-installed in the device main body is composed of, for example, a ROM in which programs are recorded, a hard disk as a storage device, or the like.

In this specification, the steps of writing a program recorded on a recording medium are not only processes that are performed chronologically in that order, but also processes that are not necessarily chronologically processed, and that are performed in parallel or individually. It also includes the processing to be performed.

Although several embodiments of the present invention have been described above, these embodiments are merely examples and do not limit the technical scope of the present invention. The present invention can take various other embodiments, and various modifications such as omissions and substitutions can be made without departing from the gist of the present invention. These embodiments and their modifications are included in the scope and gist of the invention described in this specification and the like, and are included in the scope of the invention described in the claims and equivalents thereof.

REFERENCE SIGNS LIST 10 machine learning device 40 storage battery 111 first input data acquisition unit 112 first label acquisition unit 113 first learning model construction unit

Claims (5)

an input data acquisition unit that acquires image data of at least one of the exterior and interior of the storage battery as input data;
a label acquisition unit that acquires, as a label, state information indicating the state of the storage battery corresponding to the image data;
A machine learning device comprising: a model construction unit that constructs a learning model for determining the state of the storage battery by performing supervised learning using a set of the image data and the state information as teacher data.
The label acquisition unit acquires degree information indicating a degree of abnormality as the state information,
2. The machine learning device according to claim 1, wherein the model construction unit constructs a learning model that outputs the degree information according to input image data.
The label acquisition unit acquires, as the state information, abnormality location information indicating a location of an abnormality in the storage battery corresponding to the image data,
3. The method according to claim 1 or 2, wherein the model construction unit constructs a learning model for determining the location of occurrence of the abnormality by performing supervised learning using a set of the image data and the abnormality location information as teacher data. The described machine learning device.
The input data acquisition unit acquires, as input data, specific gravity information of the electrolytic solution at a predetermined timing before the end of the life of the storage battery,
The label acquisition unit acquires, as the state information, remaining life period information indicating the remaining life of the storage battery from the predetermined timing to the end of the life of the storage battery,
4. The model building unit according to any one of claims 1 to 3, wherein the learning model for determining the remaining life is constructed by performing supervised learning using a set of the specific gravity information and the remaining life information as teacher data. The machine learning device according to
A learning model acquisition unit that acquires the learning model constructed by the machine learning device according to any one of claims 1 to 4;
a new input data acquisition unit that acquires as new input data image data of at least one of the exterior and interior of the storage battery or specific gravity information of the electrolyte of the storage battery;
and a storage battery state determination unit that determines the state of the storage battery based on the new input data and the learning model and outputs a determination result.

Images