spark开发一个程序，模拟生成新能源汽车数据集（字段信息必须包含：车架号、车型、采集时间、车速、行驶里程、故障码列表）

public class CanData implements Serializable {
    /**
     * 车架号
     */
    private String vin;
    /**
     * 车型
     */
    private String vehType;
    /**
     * 采集时间
     */
    private String collectTime;
    /**
     * 车速
     */
    private Integer speed;
    /**
     * 行驶里程
     */
    private Integer mileage;
    /**
     * 故障碍列表
     */
    private List<Integer> errCoedList;

    public String getVin() {
        return vin;
    }

    public void setVin(String vin) {
        this.vin = vin;
    }

    public String getVehType() {
        return vehType;
    }

    public void setVehType(String vehType) {
        this.vehType = vehType;
    }

    public String getCollectTime() {
        return collectTime;
    }

    public void setCollectTime(String collectTime) {
        this.collectTime = collectTime;
    }

    public Integer getSpeed() {
        return speed;
    }

    public void setSpeed(Integer speed) {
        this.speed = speed;
    }

    public Integer getMileage() {
        return mileage;
    }

    public void setMileage(Integer mileage) {
        this.mileage = mileage;
    }

    public List<Integer> getErrCoedList() {
        return errCoedList;
    }

    public void setErrCoedList(List<Integer> errCoedList) {
        this.errCoedList = errCoedList;
    }
}

2、随机生成一辆车一次行程的车辆数据代码片段

public static List<CanData> canDataListOfOneCarOneTrip(String initCollectTime, Integer driveMinutes) throws ParseException {
        List<CanData> canDataList = new ArrayList<>();

        // 根据行驶时长计算要生成的数据条数
        int totalNums = driveMinutes*60/2;

        // 生成车架号、车型
        String curVin = randomVin();
        String curVehType = randomVehType();
        // 初始化一个采集时间
        String preCollectTime = initCollectTime;

        // 循环生成车辆数据
        for(int i = 0; i < totalNums; i++) {
            CanData canData = new CanData();

            // 获取采集时间
            preCollectTime = getNewByPreCollectTime(preCollectTime);

            // 获取车速
            Integer speed = randomSpeed();

            // 获取行驶里程
            Integer mileage = randomMileage();

            // 获取故障码列表
            List<Integer> errCodeList = randomErrCode();

            // 更新canData
            canData.setVin(curVin);
            canData.setVehType(curVehType);
            canData.setCollectTime(preCollectTime);
            canData.setSpeed(speed);
            canData.setMileage(mileage);
            canData.setErrCoedList(errCodeList);

            canDataList.add(canData);
        }

        return canDataList;
    }

3数据写入HDFS代码片段

public void writeCanDataToHdfs(List<CanData> canDataList) throws IOException {
        if(canDataList.size() == 0) {
            return ;
        }

        int curHandedIndex = 0;
        // 循环写入新的文件
        while(curHandedIndex < canDataList.size()) {
            if (curLineNum >= MAX_LINE_NUMS) {
                this.curFileNameSuffix += 1;
                curLineNum = 0;
            }

            // 创建新的文件
            String curCanDataFilePath = getCurCanDataFilePath();
            FSDataOutputStream out = null;
            if(fs.exists(new Path(curCanDataFilePath))) {
                out  = fs.append(new Path(curCanDataFilePath));
            } else {
                out = fs.create(new Path(curCanDataFilePath));
            }
            BufferedWriter writer = new BufferedWriter(new OutputStreamWriter(out));
            Gson gson = new Gson();

            // 循环将数据写入HDFS
            for(; curHandedIndex < canDataList.size() && curLineNum < MAX_LINE_NUMS; curHandedIndex++) {
                CanData canData = canDataList.get(curHandedIndex);
                writer.write(gson.toJson(canData)+"\r\n");
                curLineNum ++;
            }
            writer.close();
        }
        // 特殊处理：避免追加操作。伪分布式下，append会报错。
        if(this.curLineNum > 0 && this.curLineNum < MAX_LINE_NUMS) {
            this.curFileNameSuffix ++;
            curLineNum = 0;
        }